| Let’s Get Personal: Personal Questions ImproveSocialBot Performance in the Alexa Prize | There has been an increased focus on creating conversational open-domain dialogue systems in the spoken dialogue community. Unlike traditional dialogue systems, these conversational systems cannot assume any specific information need or domain restrictions, i.e., the only inherent goal is to converse with the user on an unknown set of topics. While massive improvements in Natural Language Understanding (NLU) and the growth of available knowledge resources can partially support a robust conversation, these conversations generally lack the rapport between two humans that know each other. We developed a robust open-domain conversational system, Athena, that real Amazon Echo users access and evaluate at scale in the context of the Alexa Prize competition. We experiment with methods intended to increase intimacy between Athena and the user by heuristically developing a rule-based user model that personalizes both the current and subsequent conversations and evaluating specific personal opinion question strategies in A/B studies. Our results show a statistically significant positive impact on perceived conversation quality and length when employing these strategies. | IN SCOPE | The intention/contextualization inference for personalization/ The selection mechanism of personalized response generation / Latent information extraction from personalized response / The evaluation metric of Personalization/ Personas for the agent/Personalisation of adaptive circumstance/context | IWIDS 2023 | 2023 | personalized conversational agents |
| Dialog act guided contextual adapter for personalized speech recognition | Personalization in multi-turn dialogs has been a long standingchallenge for end-to-end automatic speech recognition (E2E ASR)models. Recent work on contextual adapters has tackled rare wordrecognition using user catalogs. This adaptation, however, does notincorporate an important cue, the dialog act, which is available ina multi-turn dialog scenario. In this work, we propose a dialog actguided contextual adapter network. Specifically, it leverages dialogacts to select the most relevant user catalogs and creates queries basedon both – the audio as well as the semantic relationship between thecarrier phrase and user catalogs to better guide the contextual biasing.On industrial voice assistant datasets, our model outperforms boththe baselines - dialog act encoder-only model, and the contextualadaptation, leading to the most improvement over the no-contextmodel: 58% average relative word error rate reduction (WERR) in themulti-turn dialog scenario, in comparison to the prior-art contextualadapter, which has achieved 39% WERR over the no-context model. | IN SCOPE | Modification of Neural network architecture/ Personalized (via Prompting) approach | ICASSP 2023 | 2023 | Adaptive dialgoue |
| You Truly Understand What I Need: Intellectual and Friendly Dialogue Agents grounding Knowledge and Persona | To build a conversational agent that interacts fluently with humans, previous studies blend knowledge or personal profile into the pre-trained language model. However, the model that considers knowledge and persona at the same time is still limited, leading to hallucination and a passive way of using personas. We propose an effective dialogue agent that grounds external knowledge and persona simultaneously. The agent selects the proper knowledge and persona to use for generating the answers with our candidate scoring implemented with a poly-encoder. Then, our model generates the utterance with lesser hallucination and more engagingness utilizing retrieval augmented generation with knowledge-persona enhanced query. We conduct experiments on the persona-knowledge chat and achieve state-of-the-art performance in grounding and generation tasks on the automatic metrics. Moreover, we validate the answers from the models regarding hallucination and engagingness through human evaluation and qualitative results. We show our retriever's effectiveness in extracting relevant documents compared to the other previous retrievers, along with the comparison of multiple candidate scoring methods. | IN SCOPE |
Modification of Neural network architecture / The selection mechanism of personalized response generation/ The evaluation metric of Personalization | EMNLP | 2022 | persona-based agent |
| Health-focused conversational agents in person-centered care: a review of apps | Health-focused apps with chatbots (“healthbots”) have a critical role in addressing gaps in quality healthcare. There is limited evidence on how such healthbots are developed and applied in practice. Our review of healthbots aims to classify types of healthbots, contexts of use, and their natural language processing capabilities. Eligible apps were those that were health-related, had an embedded text-based conversational agent, available in English, and were available for free download through the Google Play or Apple iOS store. Apps were identified using 42Matters software, a mobile app search engine. Apps were assessed using an evaluation framework addressing chatbot characteristics and natural language processing features. The review suggests uptake across 33 low- and high-income countries. Most healthbots are patient-facing, available on a mobile interface and provide a range of functions including health education and counselling support, assessment of symptoms, and assistance with tasks such as scheduling. Most of the 78 apps reviewed focus on primary care and mental health, only 6 (7.59%) had a theoretical underpinning, and 10 (12.35%) complied with health information privacy regulations. Our assessment indicated that only a few apps use machine learning and natural language processing approaches, despite such marketing claims. Most apps allowed for a finite-state input, where the dialogue is led by the system and follows a predetermined algorithm. Healthbots are potentially transformative in centering care around the user; however, they are in a nascent state of development and require further research on development, automation and adoption for a population-level health impact. | IN SCOPE | The evaluation metric of Personalization | npj Digital Medicine volume | 2022 | personalized conversational agents |
| AI-generated characters for supporting personalized learning and well-being | Advancements in machine learning have recently enabled the hyper-realistic synthesis of prose, images, audio and video data,
in what is referred to as artificial intelligence (AI)-generated media. These techniques offer novel opportunities for creating
interactions with digital portrayals of individuals that can inspire and intrigue us. AI-generated portrayals of characters can
feature synthesized faces, bodies and voices of anyone, from a fictional character to a historical figure, or even a deceased
family member. Although negative use cases of this technology have dominated the conversation so far, in this Perspective
we highlight emerging positive use cases of AI-generated characters, specifically in supporting learning and well-being. We
demonstrate an easy-to-use AI character generation pipeline to enable such outcomes and discuss ethical implications as well
as the need for including traceability to help maintain trust in the generated media. As we look towards the future, we foresee
generative media as a crucial part of the ever growing landscape of human–AI interaction. | IN SCOPE | Personas for the agent | nature machine intelligence | 2021 | personalized conversational agents |
| Discerning conversational context in online health communities for personalized digital behavior change solutions using Pragmatics to Reveal Intent in Social Media (PRISM) framework | Abstract
Background
Online health communities (OHCs) have emerged as prominent platforms for behavior modification, and the digitization of online peer interactions has afforded researchers with unique opportunities to model multilevel mechanisms that drive behavior change. Existing studies, however, have been limited by a lack of methods that allow the capture of conversational context and socio-behavioral dynamics at scale, as manifested in these digital platforms.
Objective
We develop, evaluate, and apply a novel methodological framework, Pragmatics to Reveal Intent in Social Media (PRISM), to facilitate granular characterization of peer interactions by combining multidimensional facets of human communication.
Methods
We developed and applied PRISM to analyze peer interactions (N = 2.23 million) in QuitNet, an OHC for tobacco cessation. First, we generated a labeled set of peer interactions (n = 2,005) through manual annotation along three dimensions: communication themes (CTs), behavior change techniques (BCTs), and speech acts (SAs). Second, we used deep learning models to apply our qualitative codes at scale. Third, we applied our validated model to perform a retrospective analysis. Finally, using social network analysis (SNA), we portrayed large-scale patterns and relationships among the aforementioned communication dimensions embedded in peer interactions in QuitNet.
Results
Qualitative analysis showed that the themes of social support and behavioral progress were common. The most used BCTs were feedback and monitoring and comparison of behavior, and users most commonly expressed their intentions using SAs—expressive and emotion. With additional in-domain pre-training, bidirectional encoder representations from Transformers (BERT) outperformed other deep learning models on the classification tasks. Content-specific SNA revealed that users’ engagement or abstinence status is associated with the prevalence of various categories of BCTs and SAs, which also was evident from the visualization of network structures.
Conclusions
Our study describes the interplay of multilevel characteristics of online communication and their association with individual health behaviors. | OUT OF SCOPE | | Journal of Biomedical Informatics | 2023 | personalized conversational agents |
| FedPerC: Federated Learning for Language Generation with Personal and Context Preference Embeddings | Federated learning is a training paradigm that learns from multiple distributed users without aggregat-ing data on a centralized server. Such a paradigm promises the ability to deploy machine-learningat-scale to a diverse population of end-users without first collecting a large, labeled dataset for allpossible tasks. As federated learning typically averages learning updates across a decentralized popu-lation, there is a growing need for personalization of federated learning systems (i.e conversationalagents must be able to personalize to aspecificuser’s preferences). In this work, we propose a newdirection for personalization research within federated learning, leveraging both personal embed-dings and shared context embeddings. We also present an approach to predict these “preference”embeddings, enabling personalization without backpropagation. Compared to state-of-the-art person-alization baselines, our approach achieves a 50% improvement in test-time perplexity using 0.001% ofthe memory required by baseline approaches, and achieving greater sample- and compute-efficiency | IN SCOPE | Representation learning for personalisation/ Modification of Neural network architecture / The intention/contextualisation inference for personalisation | EACL | 2022 | personalized conversational agents |
| Towards Persona-Based Empathetic Conversational Models | Empathetic conversational models have beenshown to improve user satisfaction and taskoutcomes in numerous domains. In Psychol-ogy, persona has been shown to be highly cor-related to personality, which in turn influencesempathy. In addition, our empirical analysisalso suggests that persona plays an importantrole in empathetic conversations. To this end,we propose a new task towards persona-basedempathetic conversations and present the firstempirical study on the impact of personaon empathetic responding. Specifically, wefirst present a novel large-scale multi-domaindataset for persona-based empathetic conversa-tions. We then propose CoBERT, an efficientBERT-based response selection model that ob-tains the state-of-the-art performance on ourdataset. Finally, we conduct extensive exper-iments to investigate the impact of personaon empathetic responding. Notably, our re-sults show that persona improves empatheticresponding more when CoBERT is trained onempathetic conversations than non-empatheticones, establishing an empirical link betweenpersona and empathy in human conversations. | IN SCOPE | Personas for the agent / Modification of Neural network architecture / The selection mechanism of personalized response generation | EMNLP | 2020 | personalized conversational agents |
| Will I Sound Like Me? Improving Persona Consistency in Dialogues through Pragmatic Self-Consciousness | We explore the task of improving persona consistency of dialogue agents. Recent models tackling consistency often train with additional Natural Language Inference (NLI) labels or attach trained extra modules to the generative agent for maintaining consistency. However, such additional labels and training can be demanding. Also, we find even the best-performing persona-based agents are insensitive to contradictory words. Inspired by social cognition and pragmatics, we endow existing dialogue agents with public self-consciousness on the fly through an imaginary listener. Our approach, based on the Rational Speech Acts framework (Frank and Goodman, 2012), can enforce dialogue agents to refrain from uttering contradiction. We further extend the framework by learning the distractor selection, which has been usually done manually or randomly. Results on Dialogue NLI (Welleck et al., 2019) and PersonaChat (Zhang et al., 2018) dataset show that our approach reduces contradiction and improves consistency of existing dialogue models. Moreover, we show that it can be generalized to improve context-consistency beyond persona in dialogues. | IN SCOPE | | EMNLP | 2020 | persona-based agent |
| Towards a New generation of Personalized Intelligent Conversational Agents | The Personalized Intelligent Conversational Agents workshop focuses on both long-term engaging spoken dialogue systems and text-based chatbots, as well as conversational recommender systems. The goal of the workshop is to stimulate discussion around problems, challenges, possible solutions and research directions regarding the exploitation of natural language processing and machine learning techniques to learn user features and to use them to personalize the dialogue in the next generation of intelligent conversational agents. | OUT OF SCOPE | | ACM UMAP | 2021 | personalized conversational agents |
| Hey Google, Do You Have a Personality? Designing Personality and Personas for Conversational Agents | Conversational agents designed to interact through natural language are often imbued with human-like personalities. At times,
the agent might also have a distinct persona with traits such as
gender, age, or a backstory. Designing such personality or persona
for conversational agents has become a common design practice.
In this work, we review the emerging literature on designing agent
persona or personality, and reflect on these approaches, along with
the personas that are created for common conversational agents. We
discuss open questions with regards to three aspects: meeting user
needs, the ethics of deception, and reinforcing social stereotypes
through conversational agents. We hope this work can provoke
researchers and practitioners to critically reflect on their approach
for designing personality or persona of conversational agents | IN SCOPE | Personas for the agent | ACM CUI | 2021 | personalized conversational agents |
| User-aware conversational agents | Conversational agents are becoming increasingly popular. Thesesystems present an extremely rich and challenging research spacefor addressing many aspects of user awareness and adaptation, suchas user profiles, contexts, personalities, emotions, social dynamics,conversational styles, etc. Adaptive interfaces are of long-standinginterest for the HCI community. Meanwhile, new machine learningapproaches are introduced in the current generation of conversa-tional agents, such as deep learning, reinforcement learning, andactive learning. It is imperative to consider how various aspects ofuser-awareness should be handled by these new techniques. Thegoal of this workshop is to bring together researchers in HCI, usermodeling, and the AI and NLP communities from both industry andacademia, who are interested in advancing the state-of-the-art onthe topic of user-aware conversational agents. Through a focusedand open exchange of ideas and discussions, we will work to iden-tify central research topics in user-aware conversational agents anddevelop a strong interdisciplinary foundation to address them. | OUT OF SCOPE | | ACM IUI | 2019 | personalized conversational agents |
| Labeling the Phrases of a Conversational Agent with a Unique Personalized Vocabulary | Mapping spoken text to gestures is an important research topic forrobots with conversation capabilities. According to studies on human co-speech gestures, a reasonable solution for mapping is usinga concept-based approach in which a text is first mapped to asemantic cluster (i.e., a concept) containing texts with similar meanings.Subsequently, each concept is mapped to a predefined gesture. By using a concept-based approach, this paper discusses the practical issue of obtaining concepts for a unique vocabulary personalized for a conversationalagent. UsingMicrosoft Rinna as an agent, we qualitatively compare concepts obtained automatically through a natural language processing (NLP) approach to those obtained manually through a sociological approach. Wethen identifythree limitations of the NLP approach:at the semantic level with emojis and symbols;at the semantic level with slang, new words, and buzzwords;and at the pragmaticlevel. We attribute these limitations to the personalized vocabulary of Rinna. A follow-up experiment demonstrates that robot gestures selected usinga concept-based approach leave a better impression than randomly selected gesturesfor the Rinna vocabulary, suggesting the usefulness of a concept-based gesture generation system for personalized vocabularies. Thisstudyprovides insights into the development of gesturegeneration systemsfor conversationalagentswith personalized vocabularies. | IN SCOPE | The selection mechanism of personalized response generation | IEEE SII | 2022 | personalized conversational agents |
| Human–AI collaboration enables more empathic conversations in text-based peer-to-peer mental health support | Advances in artificial intelligence (AI) are enabling systems that augment and collaborate with humans to perform simple, mechanistic tasks such as scheduling meetings and grammar-checking text. However, such human–AI collaboration poses challenges for more complex tasks, such as carrying out empathic conversations, due to the difficulties that AI systems face in navigating complex human emotions and the open-ended nature of these tasks. Here we focus on peer-to-peer mental health support, a setting in which empathy is critical for success, and examine how AI can collaborate with humans to facilitate peer empathy during textual, online supportive conversations. We develop HAILEY, an AI-in-the-loop agent that provides just-in-time feedback to help participants who provide support (peer supporters) respond more empathically to those seeking help (support seekers). We evaluate HAILEY in a non-clinical randomized controlled trial with real-world peer supporters on TalkLife (N = 300), a large online peer-to-peer support platform. We show that our human–AI collaboration approach leads to a 19.6% increase in conversational empathy between peers overall. Furthermore, we find a larger, 38.9% increase in empathy within the subsample of peer supporters who self-identify as experiencing difficulty providing support. We systematically analyse the human–AI collaboration patterns and find that peer supporters are able to use the AI feedback both directly and indirectly without becoming overly reliant on AI while reporting improved self-efficacy post-feedback. Our findings demonstrate the potential of feedback-driven, AI-in-the-loop writing systems to empower humans in open-ended, social and high-stakes tasks such as empathic conversations. | IN SCOPE | The evaluation metric of Personalization | Nature Machine Intelligence | 2023 | personalized chatbots |
| Persona-Based Conversational AI: State of the Art and Challenges | Abstract—Conversational AI has become an increasinglyprominent and practical application of machine learning. How-ever, existing conversational AI techniques still suffer from var-ious limitations. One such limitation is a lack of well-developedmethods for incorporating auxiliary information that could helpa model understand conversational context better. In this paper,we explore how persona-based information could help improvethe quality of response generation in conversations. First, weprovide a literature review focusing on the current state-of-the-artmethods that utilize persona information. We evaluate two strongbaseline methods, the Ranking Profile Memory Network andthe Poly-Encoder, on the NeurIPS ConvAI2 benchmark dataset.Our analysis elucidates the importance of incorporating personainformation into conversational systems. Additionally, our studyhighlights several limitations with current state-of-the-art meth-ods and outlines challenges and future research directions foradvancing personalized conversational AI technology | IN SCOPE | The selection mechanism of personalized response generation / Representation learning for personalization | ICDMW | 2022 | persona-based agents |
| Personalized Prompt Learning for Explainable Recommendation | Providing user-understandable explanations to justify recommendations could help users better understand the recommended items, increase the system’s ease of use, and gain users’ trust. A typical approach to realize it is natural language generation. However, previous works mostly adopt recurrent neural networks to meet the ends, leaving the potentially more effective pre-trained Transformer models under-explored. In fact, user and item IDs, as important identifiers in recommender systems, are inherently in different semantic space as words that pre-trained models were already trained on. Thus, how to effectively fuse IDs into such models becomes a critical issue. Inspired by recent advancement in prompt learning, we come up with two solutions: find alternative words to represent IDs (called discrete prompt learning) and directly input ID vectors to a pre-trained model (termed continuous prompt learning). In the latter case, ID vectors are randomly initialized but the model is trained in advance on large corpora, so they are actually in different learning stages. To bridge the gap, we further propose two training strategies: sequential tuning and recommendation as regularization. Extensive experiments show that our continuous prompt learning approach equipped with the training strategies consistently outperforms strong baselines on three datasets of explainable recommendation. | IN SCOPE | Personalized (via Prompting) approach | ACM Transactions on Information Systems | 2023 | Personalized prompt learning |
| Dialogue-adaptive language model pre-training from quality estimation | Pre-trained language models (PrLMs) have achieved great success on a wide range of natural language processing tasks by virtue of the universal language representation ability obtained by self-supervised learning on a large corpus. These models are pre-trained on standard plain texts with general language model (LM) training objectives, which would be insufficient to model dialogue-exclusive attributes like specificity and informativeness reflected in these tasks that are not explicitly captured by the pre-trained universal language representations. In this work, we propose dialogue-adaptive pre-training objectives (DAPO) derived from quality estimation to simulate dialogue-specific features, namely coherence, specificity, and informativeness. As the foundation for model pre-training, we synthesize a new dialogue corpus and build our training set with two unsupervised methods: 1) coherence-oriented context corruption, including utterance ordering, insertion, and replacement, to help the model capture the coherence inside the dialogue contexts; and 2) specificity-oriented automatic rescoring, which encourages the model to measure the quality of the synthesized data for dialogue-adaptive pre-training by considering specificity and informativeness. Experimental results on widely used open-domain response selection and quality estimation benchmarks show that DAPO significantly improves the baseline models and achieves state-of-the-art performance on the MuTual leaderboard, verifying the effectiveness of estimating quality evaluation factors into pre-training. | OUT OF SCOPE | | arxiv | 2020 | Adaptive dialgoue |
| Artificial intelligence empowered conversational agents: A systematic literature review and research agenda | Consumer research on conversational agents (CAs) has been growing. To illustrate and map out research in this field, we conducted a systematic literature review (SLR) of published work indexed in the Clarivate Web of Science and Elsevier Scopus databases. Four dominant topical areas were identified through bibliographic coupling. They are 1) consumers’ trust in CAs; 2) Natural Language Processing (NLP) in developing and designing CAs; 3) communication with CAs; 4) impact of CAs on value creation and the value of CAs for business. We leverage these findings to provide an updated synopsis of extant scientific work. Moreover, we draw a framework whereby we identify the: 1) drivers of and motivators for adoption and engagement with CAs; and 2) the outcomes of CA adoption for both users and organizations. Finally, we leverage the framework to develop an agenda for future research. | OUT OF SCOPE | | Journal of Business Research | 2019 | personalized conversational agents |
| LaMP: When Large Language Models Meet Personalization | This paper highlights the importance of per-sonalization in the current state of natural lan-guage understanding and generation and in-troduces the LaMP benchmark — a novelbenchmark for training and evaluating lan-guage models for producing personalized out-puts. LaMP offers a comprehensive evalua-tion framework with diverse language tasksand multiple entries for each user profile. Itconsists of seven personalized tasks, spanningthree classification and four text generationtasks. We also propose a retrieval augmenta-tion approach that retrieves personalized itemsfrom user profiles to construct personalizedprompts for large language models. Our base-line zero-shot and fine-tuned model results in-dicate that LMs utilizing profile augmentationoutperform their counterparts that do not fac-tor in profile information | IN SCOPE | The evaluation metric of Personalization | arxiv | 2023 | Personalized Language Generation |
| A Dual Latent Variable Personalized Dialogue Agent | Personalized dialogue agents are capable of generating responses consistent with a specific persona. Typically, personalized dialogue agents generate responses based on both the dialogue history and a representation of the agent’s desired persona. As it is impractical to obtain the persona representations for every interlocutor in real-world implementations, recent works have explored the possibility of generating personalized dialogue by finetuning the agent with dialogue examples corresponding to a given persona instead. However, in real-world implementations, a sufficient number of corresponding dialogue examples are also rarely available. Hence, in this paper, we introduce the Dual Latent Variable Generator (DLVGen), a variational personalized dialogue agent capable of generating personalized dialogue without any persona information or any corresponding dialogue examples. Unlike previous works, DLVGen models the latent distribution over potential dialogue response intents as well as the latent distribution over the agent’s potential persona. During inference, latent variables are sampled from both distributions and fed to the decoder. Extensive experiments on the popular ConvAI2 personalized dialogue corpus show that DLVGen is capable of generating natural, persona consistent responses. Additionally, we also introduce a variance regularization and response selection approach which further improved overall response quality. | IN SCOPE | The data augmentation of personalisation / The selection mechanism of personalized response generation/Latent information extraction from personalized response | SN Computer Science | 2023 | personalized conversational agents |
| Pre-Evaluation with a Personalized Feedback Conversational Agent Integrated in Moodle | Abstract—Pre-evaluation of the learner's level is a common learning strategy
designed to determine the prior knowledge and skills of learners. A pre-evaluation is carried out at the beginning of the course and based on the results obtained,
personalized resources will be provided that respond to individual learner needs.
This paper presents a pre-evaluation for a C programming language course by
providing, at the end of the quiz, a personalized formative feedback and recommendation to the learners. We have developed our conversational chatbot named
QuizCbot, which allows learners to go directly to the parts where they need the
most help through the personalized feedback provided to them, including their
final scores, the questions they answered correctly and the questions they answered incorrectly with the correct answer and explanation. Hence, the chatbot
makes a recommendation on the concepts in which the learner did not obtain the
average, identifying the concepts not mastered where the learner needs more (or
less) support. Determining what learners know and don't know can help to improve the learning experience. We have integrated our QuizCbot chatbot, which
is based on Natural Language Understanding (NLU), into the Moodle learning
environment | IN SCOPE | The evaluation metric of Personalization | International Journal of Emerging Technologies in Learning | 2023 | personalized conversational agents |
| Personalized Search-based Query Rewrite System for Conversational AI | Query rewrite (QR) is an emerging compo-nent in conversational AI systems, reducinguser defect. User defect is caused by vari-ous reasons, such as errors in the spoken dia-logue system, users’ slips of the tongue or theirabridged language. Many of the user defectsstem from personalized factors, such as user’sspeech pattern, dialect, or preferences. In thiswork, we propose a personalized search-basedQR framework, which focuses on automaticreduction of user defect. We build a person-alized index for each user, which encompassesdiverse affinity layers to reflect personal pref-erences for each user in the conversational AI.Our personalized QR system contains retrievaland ranking layers. Supported by user feed-back based learning, training our models doesnot require hand-annotated data. Experimentson personalized test set showed that our per-sonalized QR system is able to correct system-atic and user errors by utilizing phonetic andsemantic inputs | IN SCOPE | Persona data Distillation and downsampling / The selection mechanism of personalized response generation | Proceedings of the 3rd Workshop on Natural Language Processing for Conversational AI | 2021 | personalized conversational agents |
| PAIGE: Personalized Adaptive Interactions Graph Encoderfor Query Rewriting in Dialogue Systems | Unexpected responses or repeated clarificationquestions from conversational agents detractfrom the users’ experience with technologymeant to streamline their daily tasks. To reducethese frictions, Query Rewriting (QR) tech-niques replace transcripts of faulty queries withalternatives that lead to responses that satisfythe users’ needs. Despite their successes, exist-ing QR approaches are limited in their ability tofix queries that require considering users’ per-sonal preferences. We improveQRby propos-ingPersonalizedAdaptiveInteractionsGraphEncoder (PAIGE). PAIGE is the firstQRar-chitecture that jointly models user’s affinitiesand query semantics end-to-end. The core ideais to represent previous user-agent interactionsand world knowledge in a structured form —a heterogeneous graph — and apply messagepassing to propagate latent representations ofusers’ affinities to refine utterance embeddings.Using these embeddings, PAIGE can poten-tially provide different rewrites given the samequery for users with different preferences. Ourmodel, trained without any human-annotateddata, improves the rewrite retrieval precision ofstate-of-the-art baselines by 12.5–17.5% whilehaving nearly ten times fewer parameters. | IN SCOPE | Persona data Distillation and downsampling / The selection mechanism of personalized response generation / Representation learning for personalization | EMNLP | 2022 | personalized conversational agents |
| Knowledge transfer between speakers for personalised dialogue management | Model-free reinforcement learning hasbeen shown to be a promising data drivenapproach for automatic dialogue policyoptimization, but a relatively large amountof dialogue interactions is needed be-fore the system reaches reasonable perfor-mance. Recently, Gaussian process basedreinforcement learning methods have beenshown to reduce the number of dialoguesneeded to reach optimal performance, andpre-training the policy with data gatheredfrom different dialogue systems has fur-ther reduced this amount. Following thisidea, a dialogue system designed for a sin-gle speaker can be initialised with datafrom other speakers, but if the dynamics ofthe speakers are very different the modelwill have a poor performance. When datagathered from different speakers is avail-able, selecting the data from the most sim-ilar ones might improve the performance.We propose a method which automaticallyselects the data to transfer by defining asimilarity measure between speakers, anduses this measure to weight the influenceof the data from each speaker in the pol-icy model. The methods are tested by sim-ulating users with different severities ofdysarthria interacting with a voice enabledenvironmental control system. | OUT OF SCOPE | | SIGDIAL | 2015 | Personalized dialogue |
| Making conversations with chatbots more personalized | Many of the world’s leading brands and increasingly government agencies are using intelligent agent technologies, also known as chatbots to interact with consumers. However, consumer satisfaction with chatbots is
mixed. Consumers report frustration with chatbots arising from misunderstood questions, irrelevant responses,
and poor integration with human service agents. This study examines whether human-computer interactions can
be more personalized by matching consumer personality with congruent machine personality using language.
Although the idea that personality is manifested through language, and that people are more likely to be
responsive to others with the same personality is well known, there is a dearth of research that examines whether
this is consistent for human-computer interactions. Based on a sample of over 57,000 chatbot interactions, this
study demonstrates that consumer personality can be predicted during contextual interactions, and that chatbots
can be manipulated to ‘assume a personality’ using response language. Matching consumer personality with
congruent chatbot personality had a positive impact on consumer engagement with chatbots and purchasing
outcomes for interactions involving social gain. | IN SCOPE | Personas for the agent | Elsevier:Computers in Human Behavior | 2021 | personalized chatbots |
| Personalized Prompt for Sequential Recommendation | Pre-training models have shown their power in sequential recommendation. Recently, prompt has been widely explored andverified for tuning after pre-training in NLP, which helps to more effectively and parameter-efficiently extract useful knowledge frompre-training models for downstream tasks, especially in cold-start scenarios. However, it is challenging to bring prompt-tuning from NLPto recommendation, since the tokens of recommendation (i.e., items) are million-level and do not have concrete explainable semantics,and the sequence modeling in recommendation should be personalized. In this work, we first introduce prompt to pre-trainedrecommendation models and propose a novel Personalized prompt-based recommendation (PPR) framework for cold-startrecommendation. Specifically, we build personalized soft prompt via a prompt generator based on user profiles, and enable a sufficienttraining on prompts via a new prompt-oriented contrastive learning. PPR is effective, parameter-efficient, and universal in varioustaskss. In both few-shot and zero-shot recommendation tasks, PPR models achieve significant improvements over baselines in threelarge-scale datasets. We also verify PPR’s universality on different pre-trained recommendation models. Finally, we explore andconfirm the capability of PPR on other tasks such as cross-domain recommendation and user profile prediction, shedding lights on thepromising future directions of better using large-scale pre-trained recommendation models | IN SCOPE | Personalized (via Prompting) approach | arxiv | 2023 | Personalized prompt learning |
| MetaPrompting: Learning to Learn Better Prompts | Prompting method is regarded as one of the crucial progress for few-shot nature language processing. Recent research on prompting moves from discrete tokens based ``hard prompts'' to continuous ``soft prompts'', which employ learnable vectors as pseudo prompt tokens and achieve better performance. Though showing promising prospects, these soft-prompting methods are observed to rely heavily on good initialization to take effect. Unfortunately, obtaining a perfect initialization for soft prompts requires understanding of inner language models working and elaborate design, which is no easy task and has to restart from scratch for each new task. To remedy this, we propose a generalized soft prompting method called MetaPrompting, which adopts the well-recognized model-agnostic meta-learning algorithm to automatically find better prompt initialization that facilitates fast adaptation to new prompting tasks.Extensive experiments show MetaPrompting tackles soft prompt initialization problem and brings significant improvement on four different datasets (over 6 points improvement in accuracy for 1-shot setting), achieving new state-of-the-art performance. | IN SCOPE | Personalized (via Prompting) approach | Coling | 2022 | Personalized prompt learning |
| Response Generation with Context-Aware Prompt Learning | Pre-trained language models (PLM) havemarked a huge leap in neural dialogue mod-eling. While PLMs are pre-trained on large-scale text corpora, they are usually fine-tunedon scarce dialogue data with specific domainknowledge and dialogue styles. However, tai-loring the language models while fully utiliz-ing prior knowledge in large pre-trained mod-els remains a challenge. In this paper, wepresent a novel approach for pre-trained dia-logue modeling that casts the dialogue gener-ation problem as a prompt-learning task. In-stead of fine-tuning on limited dialogue data,our approach, DialogPrompt, learns contin-uous prompt embeddings optimized for di-alogue contexts, which appropriately elicitknowledge from the large pre-trained model.To encourage the model to better utilize theprompt embeddings, the prompt encodings aredesigned to be dynamically generated basedon the input dialogue context. Experimentson popular conversation datasets show thatour approach significantly outperforms thefine-tuning baseline and the generic prompt-learning methods. Furthermore, human evalu-ations strongly support the superiority of Di-alogPrompt in regard to response generationquality | IN SCOPE | Personalized (via Prompting) approach / Representation learning for personalization
/ Modification of Neural network architecture / The selection mechanism of personalized response generation | arxiv | 2021 | Personalized prompt dialogue |
| Contextual Dynamic Prompting for Response Generation in Task-oriented Dialog Systems | Response generation is one of the critical components in task-oriented dialog systems. Existing studies have shown that large pre-trained language models can be adapted to this task. The typical paradigm of adapting such extremely large language models would be by fine-tuning on the downstream tasks which is not only time-consuming but also involves significant resources and access to fine-tuning data. Prompting (Schick and Schütze, 2020) has been an alternative to fine-tuning in many NLP tasks. In our work, we explore the idea of using prompting for response generation in task-oriented dialog systems. Specifically, we propose an approach that performs contextual dynamic prompting where the prompts are learnt from dialog contexts. We aim to distill useful prompting signals from the dialog context. On experiments with MultiWOZ 2.2 dataset (Zang et al., 2020), we show that contextual dynamic prompts improve response generation in terms of combined score (Mehri et al., 2019) by 3 absolute points, and a massive 20 points when dialog states are incorporated. Furthermore, human annotation on these conversations found that agents which incorporate context were preferred over agents with vanilla prefix-tuning. | IN SCOPE | Personalized (via Prompting) approach / Representation learning for personalization
/ Persona data Distillation and downsampling / The selection mechanism of personalized response generation / The intention/contextualization inference for personalization | EACL | 2023 | Personalized prompt dialogue |
| Commonsense-Aware Prompting for Controllable Empathetic Dialogue Generation | Improving the emotional awareness of pre-trained language models is an emerging important problem for dialogue generation tasks. Although prior studies have introduced methods to improve empathetic dialogue generation, few have discussed how to incorporate commonsense knowledge into pre-trained language models for controllable dialogue generation. In this study, we propose a novel framework that improves empathetic dialogue generation using pre-trained language models by 1) incorporating commonsense knowledge through prompt verbalization, and 2) controlling dialogue generation using a strategy-driven future discriminator. We conducted experiments to reveal that both the incorporation of social commonsense knowledge and enforcement of control over generation help to improve generation performance. Finally, we discuss the implications of our study for future research. | IN SCOPE | Personalized (via Prompting) approach / The selection mechanism of personalized response generation / Modification of Neural network architecture
/ Representation learning for personalization | AAAI | 2023 | Personalized prompt dialogue |
| A Dual Prompt Learning Framework for Few-Shot Dialogue State Tracking | Dialogue state tracking (DST) module is an important component for task-oriented dialog systems to understand users' goals and needs. Collecting dialogue state labels including slots and values can be costly, especially with the wide application of dialogue systems in more and more new-rising domains. In this paper, we focus on how to utilize the language understanding and generation ability of pre-trained language models for DST. We design a dual prompt learning framework for few-shot DST. Specifically, we consider the learning of slot generation and value generation as dual tasks, and two prompts are designed based on such a dual structure to incorporate task-related knowledge of these two tasks respectively. In this way, the DST task can be formulated as a language modeling task efficiently under few-shot settings. Experimental results on two task-oriented dialogue datasets show that the proposed method not only outperforms existing state-of-the-art few-shot methods, but also can generate unseen slots. It indicates that DST-related knowledge can be probed from PLM and utilized to address low-resource DST efficiently with the help of prompt learning. | IN SCOPE | The data augmentation of personalisation / Personalized (via Prompting) approach | WWW | 2023 | Personalized prompt dialogue |
| A Stack-Propagation Framework for Low-Resource Personalized Dialogue Generation | With the resurgent interest in building open-domain dialogue systems, the dialogue generation task has attracted increasing attention over the past few years. This task is usually formulated as a conditional generation problem, which aims to generate a natural and meaningful response given dialogue contexts and specific constraints, such as persona. And maintaining a consistent persona is essential for the dialogue systems to gain trust from the users. Although tremendous advancements have been brought, traditional persona-based dialogue models are typically trained by leveraging a large number of persona-dense dialogue examples. Yet, such persona-dense training data are expensive to obtain, leading to a limited scale. This work presents a novel approach to learning from limited training examples by regarding consistency understanding as a regularization of response generation. To this end, we propose a novel stack-propagation framework for learning a generation and understanding pipeline. Specifically, the framework stacks a Transformer encoder and two Transformer decoders, where the first decoder models response generation and the second serves as a regularizer and jointly models response generation and consistency understanding. The proposed framework can benefit from the stacked encoder and decoders to learn from much smaller personalized dialogue data while maintaining competitive performance. Under different low-resource settings, subjective and objective evaluations prove that the stack-propagation framework outperforms strong baselines in response quality and persona consistency and largely overcomes the shortcomings of traditional models that rely heavily on the persona-dense dialogue data. | IN SCOPE | Personas for the agent / Modification of Neural network architecture / The selection mechanism of personalized response generation / The evaluation metric of Personalization | ACM Transactions on Information Systems | 2023 | Personalized prompt dialogue |
| pFedPrompt: Learning Personalized Prompt for Vision-Language Models in Federated Learning | Pre-trainedvision-languagemodelslikeCLIPshowgreatpotentialinlearningrepresentationsthatcapturelatentcharacteristicsofusers.ArecentlyproposedmethodcalledContextualOptimization(CoOp)introducestheconceptoftrainingpromptforadaptingpre-trainedvision-languagemodels.Giventhelightweightnatureofthismethod,researchershavemigratedtheparadigmfromcentral-izedtodecentralizedsystemtoinnovatethecollaborativetrainingframeworkofFederatedLearning(FL).However,currentprompttraininginFLmainlyfocusesonmodelinguserconsensusandlackstheadaptationtousercharacteristics,leavingthepersonalizationofpromptlargelyunder-explored.ResearchesoverthepastfewyearshaveappliedpersonalizedFL(pFL)approachestocustomiz-ingmodelsforheterogeneoususers.Unfortunately,wefndthatwiththevariationofmodalityandtrainingbehavior,directlyap-plyingthepFLmethodstoprompttrainingleadstoinsufcientpersonalizationandperformance.Tobridgethegap,wepresentpFedPrompt,whichleveragestheuniqueadvantageofmultimodal-ityinvision-languagemodelsbylearninguserconsensusfromlinguisticspaceandadaptingtousercharacteristicsinvisualspaceinanon-parametricmanner.Throughthisdualcollaboration,thelearnedpromptwillbefullypersonalizedandalignedtotheuser’slocalcharacteristics.Weconductextensiveexperimentsacrossvar-iousdatasetsundertheFLsettingwithstatisticalheterogeneity.TheresultsdemonstratethesuperiorityofourpFedPromptagainstthealternativeapproacheswithrobustperformance. | OUT OF SCOPE | | WWW | 2023 | Personalized prompt dialogue |
| Linguistic Features to Consider When Applying Persona of the Real Person to the Text-based Agent | As artificial intelligence (AI) technologies advance, the possibility of developing virtual agents capable of mimicking human beings is increasing. Furthermore, AI techniques applicable to mimicking certain features of a specific person (e.g., facial expression, voice, motion) are becoming more sophisticated. Although the HCI community has explored how to design or develop AI agents mimicking a real person, limited studies on mimicking someone’s text-based behavior shown in the instant messaging exist. This study investigates the features that make users perceive text-based agents as people they know in reality. On top of the previous efforts of designing human-like virtual agents, our work suggests design guidelines for applying the persona of the real person (PRP) to text-based agents. | IN SCOPE | Personas for the agent | MobileHCI | 2020 | persona-based agent |
| Towards Personalized User Interface Design For News Chatbots: A Pilot Study | Numerous mainstream news organizations have adopted chatbots as an emerging channel for delivering personalized news services. However, the news presentation style of chatbots may not satisfy users’ personalization needs due to the “one-size-fits-all” design. In this context, we conducted two within-subject studies to investigate the impacts of news reader types, namely “dipper”, “tracker”, and “reviewer” on users’ preference for different interfaces for news selection and reading. Our preliminary findings reveal that all these types of news readers perceived a higher level of ease of selection when using the carousel design. For “trackers” and “reviewers”, the question-driven design resulted in a higher level of perceived ease of reading than the in-conversation design. Finally, I discuss my findings and provide insights into personalized user interface design for news chatbots. | IN SCOPE | The selection mechanism of personalized response generation | CHI | 2023 | personalized chatbots |
| Personalized Chatbot Trustworthiness Ratings | Abstract—In this paper, we address a setting where a conver-sation agent, also know as a chatbot, cannot be modified and itstraining data cannot be accessed, and yet a neutral party wantsto assess and communicate its trustworthiness to a user in away that is tailored to the user’s priorities over the various trustissues (such as bias, abusive language, information leakage, orinappropriate communication complexity). Such a rating can helpusers choose among alternative chatbots, developers test theirsystems, business leaders price their offerings, and regulators setpolicies. We describe a chatbot rating methodology that relieson separate rating modules for each trust issue, and on users’priority orderings among the relevant trust issues, to generatean aggregate personalized rating for the trustworthiness of achatbot. The method is independent of the specific trust issuesand is parametric to the aggregation procedure, thereby allowingfor seamless generalization. We illustrate its general use, integrateit with a live chatbot, and evaluate it on four dialog datasets andrepresentative user profiles, validated with a user survey | IN SCOPE | The selection mechanism of personalized response generation | arxiv | 2020 | personalized chatbots |
| Personalized Quest and Dialogue Generation in Role-Playing Games: A Knowledge Graph- and Language Model-based Approach | Procedural content generation (PCG) in video games offers unprecedented opportunities for customization and user engagement. Working within the specialized context of role-playing games (RPGs), we introduce a novel framework for quest and dialogue generation that places the player at the core of the generative process. Drawing on a hand-crafted knowledge base, our method grounds generated content with in-game context while simultaneously employing a large-scale language model to create fluent, unique, accompanying dialogue. Through human evaluation, we confirm that quests generated using this method can approach the performance of hand-crafted quests in terms of fluency, coherence, novelty, and creativity; demonstrate the enhancement to the player experience provided by greater dynamism; and provide a novel, automated metric for the relevance between quest and dialogue. We view our contribution as a critical step toward dynamic, co-creative narrative frameworks in which humans and AI systems jointly collaborate to create unique and user-specific playable experiences. | IN SCOPE | The selection mechanism of personalized response generation / Personas for the agent | CHI | 2023 | Personalized Language Generation |
| Personalized Chit-Chat Generation for Recommendation Using External Chat Corpora | Chit-chat has been shown effective in engaging users in human-computer interaction. We find with a user study that generating appropriate chit-chat for news articles can help expand user interest and increase the probability that a user reads a recommended news article. Based on this observation, we propose a method to generate personalized chit-chat for news recommendation. Different from existing methods for personalized text generation, our method only requires an external chat corpus obtained from an online forum, which can be disconnected from the recommendation dataset from both the user and item (news) perspectives. This is achieved by designing a weak supervision method for estimating users' personalized interest in a chit-chat post by transferring knowledge learned by a news recommendation model. Based on the method for estimating user interest, a reinforcement learning framework is proposed to generate personalized chit-chat. Extensive experiments, including the automatic offline evaluation and user studies, demonstrate the effectiveness of our method. | IN SCOPE | The selection mechanism of personalized response generation / The evaluation metric of Personalization / Modification of Neural network architecture / Personas for the agent | KDD | 2022 | Personalized Language Generation |
| Towards Knowledge-Based Personalized Product Description Generation in E-commerce | Quality product descriptions are critical for providing competitive customer experience in an e-commerce platform. An accurate and attractive description not only helps customers make an informed decision but also improves the likelihood of purchase. However, crafting a successful product description is tedious and highly time-consuming. Due to its importance, automating the product description generation has attracted considerable interests from both research and industrial communities. Existing methods mainly use templates or statistical methods, and their performance could be rather limited.
In this paper, we explore a new way to generate the personalized product description by combining the power of neural networks and knowledge base. Specifically, we propose a KnOwledge Based pErsonalized (or KOBE) product description generation model in the context of E-commerce. In KOBE, we extend the encoder-decoder framework, the Transformer, to a sequence modeling formulation using self-attention. In order to make the description both informative and personalized, KOBE considers a variety of important factors during text generation, including product aspects, user categories, and knowledge base, etc. Experiments on real-world datasets demonstrate that the proposed method out-performs the baseline on various metrics. KOBE can achieve an improvement of 9.7% over state-of-the-arts in terms of BLEU. We also present several case studies as the anecdotal evidence to further prove the effectiveness of the proposed approach. The framework has been deployed in Taobao, the largest online e-commerce platform in China. | OUT OF SCOPE | | KDD | 2019 | Personalized Language Generation |
| Building User-oriented Personalized Machine Translator based on User-Generated Textual Content | Machine Translation (MT) has been a very useful tool to assist multilingual communication and collaboration. In recent years, by taking advantage of the exciting developments of neural networks and deep learning, the accuracy and speed of machine translation have been continuously improved. However, most machine translation methods and systems are data-driven. They tend to select a consensus response represented in training data, while a user's preferred linguistic style, which is important for translation comprehension and user experience, is ignored. For this problem, we aim to build a user-oriented personalized machine translation model in this paper. The model aims to learn each user's linguistic style from the textual content that is generated by her/him (User-Generated Textual Content, UGTC) in social media context and generate personalized translation results utilizing several state-of-the-art deep learning techniques like Transformer and pre-training. We also implemented a user-oriented personalized machine translator using Weibo as a case of the source of UGTC to provide a systematical implementation scheme of a user-oriented personalized machine translation system based on our model. The translator was evaluated by automatic evaluation in combination with human evaluation. The results suggest that our model can generate more personalized, natural and lively translation results and enhance the comprehensibility of translation results, which makes its generations more preferred by users versus general translation results. | OUT OF SCOPE | | Proceedings of the ACM on Human-Computer Interaction | 2022 | Personalized Language Generation |
| Personalized Response Generation via Domain adaptation | In this paper, we propose a novel personalized response generation model via domain adaptation (PRG-DM). First, we learn the human responding style from large general data (without user-specific information). Second, we fine tune the model on a small size of personalized data to generate personalized responses with a dual learning mechanism. Moreover, we propose three new rewards to characterize good conversations that are personalized, informative and grammatical. We employ the policy gradient method to generate highly rewarded responses. Experimental results show that our model can generate better personalized responses for different users. | IN SCOPE | The selection mechanism of personalized response generation / Personalisation of adaptive circumstance/context
| SIGIR | 2017 | Personalized Language Generation |
| Controlling Personality Style in Dialogue with Zero-Shot Prompt-Based Learning | rompt-based or in-context learning has achieved high zero-shot performance on many natural language generation (NLG) tasks. Here we explore the performance of prompt-based learning for simultaneously controlling the personality and the semantic accuracy of an NLG for task-oriented dialogue. We experiment with prompt-based learning on the PERSONAGE restaurant recommendation corpus to generate semantically and stylistically-controlled text for 5 different Big-5 personality types: agreeable, disagreeable, conscientious, unconscientious, and extravert. We test two different classes of discrete prompts to generate utterances for a particular personality style: (1) prompts that demonstrate generating directly from a meaning representation that includes a personality specification; and (2) prompts that rely on first converting the meaning representation to a textual pseudo-reference, and then using the pseudo-reference in a textual style transfer (TST) prompt. In each case, we show that we can vastly improve performance by over-generating outputs and ranking them, testing several ranking functions based on automatic metrics for semantic accuracy, personality-match, and fluency. We also test whether NLG personality demonstrations from the restaurant domain can be used with meaning representations for the video game domain to generate personality stylized utterances about video games. Our findings show that the TST prompts produces the highest semantic accuracy (78.46% for restaurants and 87.6% for video games) and personality accuracy (100% for restaurants and 97% for video games). Our results on transferring personality style to video game utterances are surprisingly good. To our knowledge, there is no previous work testing the application of prompt-based learning to simultaneously | IN SCOPE | Personalized (via Prompting) approach / Personas for the agent/ Representation learning for personalization / The selection mechanism of personalized response generation / The evaluation metric of Personalization | IWSDS | 2023 | prompt personalization |
| Question Personalization in an Intelligent Tutoring System | Abstract.This paper investigates personalization in the field of intelli-gent tutoring systems (ITS). We hypothesize that personalization in theway questions are asked improves student learning outcomes. Previouswork on dialogue-based ITS personalization has yet to address questionphrasing. We show that generating versions of the questions suitablefor students at different levels of subject proficiency improves studentlearning gains, using variants written by a domain expert and an experi-mental A/B test. This insight demonstrates that the linguistic realizationof questions in an ITS affects the learning outcomes for students | IN SCOPE | Personalization of adaptive circumstance/context / The evaluation metric of Personalization | AIED | 2022 | prompt personalization |
| Prompt-based System for Personality and Interpersonal Reactivity Prediction | This paper describes our proposed method for the Workshop on Computational Approaches to Subjectivity,
Sentiment & Social Media Analysis (WASSA) 2022 shared task on Personality Prediction (PER) and Reactivity
Index Prediction (IRI). In this paper, we adopt the prompt-based learning method with the pre-trained language
model to accomplish these tasks. Specifically, the prompt is designed to provide knowledge of the extra
personalized information for enhancing the pre-trained model. Data augmentation and model ensemble are
adopted for obtaining better results. Moreover, we also provided the online software demonstration and the
codes of the software for further research | OUT OF SCOPE | | Software impact | 2022 | prompt personalization |
| Sometimes You Want to Go Where Everybody Knows your Name | We introduce a new metric for measuring how well a model personalizes to a user's specific preferences. We define personalization as a weighting between performance on user specific data and performance on a more general global dataset that represents many different users. This global term serves as a form of regularization that forces us to not overfit to individual users who have small amounts of data. In order to protect user privacy, we add the constraint that we may not centralize or share user data. We also contribute a simple experiment in which we simulate classifying sentiment for users with very distinct vocabularies. This experiment functions as an example of the tension between doing well globally on all users, and doing well on any specific individual user. It also provides a concrete example of how to employ our new metric to help reason about and resolve this tension. We hope this work can help frame and ground future work into personalization. | OUT OF SCOPE | | arxiv | 2018 | prompt personalization |
| Tell Me About Yourself: Using an AI-Powered Chatbot to Conduct Conversational Surveys with Open-ended Questions | The rise of increasingly more powerful chatbots offers a new way to collect information through conversational surveys, where a chatbot asks open-ended questions, interprets a user’s free-text responses, and probes answers whenever needed. To investigate the effectiveness and limitations of such a chatbot in conducting surveys, we conducted a field study involving about 600 participants. In this study with mostly open-ended questions, half of the participants took a typical online survey on Qualtrics and the other half interacted with an AI-powered chatbot to complete a conversational survey. Our detailed analysis of over 5,200 free-text responses revealed that the chatbot drove a significantly higher level of participant engagement and elicited significantly better quality responses measured by Gricean Maxims in terms of their informativeness, relevance, specificity, and clarity. Based on our results, we discuss design implications for creating AI-powered chatbots to conduct effective surveys and beyond. | OUT OF SCOPE | | arxiv | 2020 | prompt personalized chatbot |
| CloneBot: Personalized Dialogue-Response Predictions | Our project task was to create a model that, given a speaker ID, chat history, andan utterance query, can predict the response utterance in a conversation. Themodel is personalized for each speaker. This task can be a useful tool for buildingspeech bots that talk in a human-like manner in a live conversation. Further, wesucceeded at using dense-vector encoding clustering to be able to retrieve relevanthistorical dialogue context, a useful strategy for overcoming the input limitationsof neural-based models when predictions require longer-term references from thedialogue history. In this paper, we have implemented a state-of-the-art model usingpre-training and fine-tuning techniques built on transformer architecture and multi-headed attention blocks for the Switchboard corpus. We also show how efficientvector clustering algorithms can be used for real-time utterance predictions thatrequire no training and therefore work on offline and encrypted message histories | IN SCOPE | The selection mechanism of personalized response generation / Representation learning for personalization / Personalization of adaptive circumstance/context | arxiv | 2021 | prompt personalized chatbot |
| Learning Implicit User Profiles for Personalized Retrieval-Based Chatbot | In this paper, we explore the problem of developing personalized chatbots. A personalized chatbot is designed as a digital chatting assistant for a user. The key characteristic of a personalized chatbot is that it should have a consistent personality with the corresponding user. It can talk the same way as the user when it is delegated to respond to others' messages. We present a retrieval-based personalized chatbot model, namely IMPChat, to learn an implicit user profile from the user's dialogue history. We argue that the implicit user profile is superior to the explicit user profile regarding accessibility and flexibility. IMPChat aims to learn an implicit user profile through modeling user's personalized language style and personalized preferences separately. To learn a user's personalized language style, we elaborately build language models from shallow to deep using the user's historical responses; To model a user's personalized preferences, we explore the conditional relations underneath each post-response pair of the user. The personalized preferences are dynamic and context-aware: we assign higher weights to those historical pairs that are topically related to the current query when aggregating the personalized preferences. We match each response candidate with the personalized language style and personalized preference, respectively, and fuse the two matching signals to determine the final ranking score. Comprehensive experiments on two large datasets show that our method outperforms all baseline models. | IN SCOPE | Latent information extraction from personalized response | CIKM | 2021 | prompt personalized chatbot |
| Assigning personality/identity to a chatting machine for coherent conversation generation | Endowing a chatbot with personality or an identity is quite challenging but critical to deliver more realistic and natural conversations. In this paper, we address the issue of generating responses that are coherent to a pre-specified agent profile. We design a model consisting of three modules: a profile detector to decide whether a post should be responded using the profile and which key should be addressed, a bidirectional decoder to generate responses forward and backward starting from a selected profile value, and a position detector that predicts a word position from which decoding should start given a selected profile value. We show that general conversation data from social media can be used to generate profile-coherent responses. Manual and automatic evaluation shows that our model can deliver more coherent, natural, and diversified responses. | IN SCOPE | The selection mechanism of personalized response generation / Modification of Neural network architecture / Personas for the agent | arxiv | 2017 | prompt personalized chatbot |
| Let's Get Personal: Personal Questions Improve SocialBot Performance in the Alexa Prize | There has been an increased focus on creating conversational open-domain dialogue systems in the spoken dialogue community. Unlike traditional dialogue systems, these conversational systems cannot assume any specific information need or domain restrictions, i.e., the only inherent goal is to converse with the user on an unknown set of topics. While massive improvements in Natural Language Understanding (NLU) and the growth of available knowledge resources can partially support a robust conversation, these conversations generally lack the rapport between two humans that know each other. We developed a robust open-domain conversational system, Athena, that real Amazon Echo users access and evaluate at scale in the context of the Alexa Prize competition. We experiment with methods intended to increase intimacy between Athena and the user by heuristically developing a rule-based user model that personalizes both the current and subsequent conversations and evaluating specific personal opinion question strategies in A/B studies. Our results show a statistically significant positive impact on perceived conversation quality and length when employing these strategies. | IN SCOPE | Repeatative research topic | IWSDS | 2023 | prompt personalized chatbot |
| POSGen: Personalized Opening Sentence Generation for Online Insurance Sales | The insurance industry is shifting their sales mode from offline to online, in expectation to reach massive potential customers in the digitization era. Due to the complexity and the nature of insurance products, a cost-effective online sales solution is to exploit chatbot AI to raise customers' attention and pass those with interests to human agents for further sales. For high response and conversion rates of customers, it is crucial for the chatbot to initiate a conversation with personalized opening sentences, which are generated with user-specific topic selection and ordering. Such personalized opening sentence generation is challenging because (i) there are limited historical samples for conversation topic recommendation in online insurance sales and (ii) existing text generation schemes often fail to support customized topic ordering based on user preferences. We design POSGen, a personalized opening sentence generation scheme dedicated for online insurance sales. It transfers user embeddings learned from auxiliary online user behaviours to enhance conversation topic recommendation, and exploits a context management unit to arrange the recommended topics in user-specific ordering for opening sentence generation. POSGen is deployed on a real-world online insurance platform. It achieves 2.33x total insurance premium improvement through a two-month global test. | IN SCOPE | The selection mechanism of personalized response generation / Representation learning for personalization / Modification of Neural network architecture / The evaluation metric of Personalization / Personalization of adaptive circumstance/context | IEEE International Conference on Big Data | IEEE International Conference on Big Data | prompt personalized agent |
| PEDM: A Multi-task Learning Model for Persona-aware Emoji-embedded Dialogue Generation | As a vivid and linguistic symbol, Emojis have become a prevailing medium interspersed in text-based communication (e.g., social media and chit-chat) to express emotions, attitudes, and situations. Generally speaking, a social-oriented chatbot that can generate appropriate Emoji-embedded responses would be much more competitive, making communications more fun, engaging, and human-like. However, the current Emoji-related research is still in its infancy, leading to an awkward situation of data deficiency. How to develop an Emoji-embedded dialogue system while addressing the lack of data will be interesting and meaningful for the application of future AI. To bridge this gap, we propose a multi-task learning method for persona-aware Emoji-embedded dialogue generation in this article. Specifically, as the benchmark of model training and evaluation, which includes 1.2 million Emoji-embedded tweets and 1.1 million post-response pairs, we first construct a dataset named EmojiTweet to handle the data deficiency problem. Then, a Seq2Seq-based model with multi-task learning is designed to simultaneously learn response generation and Emoji embedding from the constructed non-Emoji dialogue and Emoji-embedded monologue data. Afterward, we incorporate persona factors into our model by adopting persona fusion and personalized bias methods to deliver personalized dialogues with more accurately selected Emojis. Finally, we conduct extensive experiments, where the experimental results and evaluations demonstrate that our model has three key benefits: improved dialogue quality, higher user engagement, and not relying on large-scale Emoji-embedded dialogue data representing specific personas. | IN SCOPE | Representation learning for personalization / The selection mechanism of personalized response generation / Personas for the agent / Modification of Neural network architecture / The evaluation metric of Personalization | ACM Transactions on Multimedia Computing, Communications, and Applications | 2023 | Personalized Dialogue Generation |
| Bilateral personalized dialogue generation with contrastive learning | Generating personalized responses is one of the major challenges in natural human-robot interaction. Current researches in this field mainly focus on generating responses consistent with the robot's pre-assigned persona, while ignoring the user's persona. Such responses may be inappropriate or even offensive, which may lead to the bad user experience. Therefore, we propose a Bilateral Personalized Dialogue Generation (BPDG) method for dyadic conversation, which integrates user and robot personas into dialogue generation via designing a dynamic persona-aware fusion method. To bridge the gap between the learning objective function and evaluation metrics, the Conditional Mutual Information Maximum (CMIM) criterion is adopted with contrastive learning to select the proper response from the generated candidates. Moreover, a bilateral persona accuracy metric is designed to measure the degree of bilateral personalization. Experimental results demonstrate that, compared with several state-of-the-art methods, the final results of the proposed method are more personalized and consistent with bilateral personas in terms of both automatic and manual evaluations. | IN SCOPE | Repeatative research topic | Soft Computing | 2023 | Personalized Dialogue Generation |
| Conversation and recommendation: knowledge-enhanced personalized dialog system | Traditional recommender systems are usually single-shot systems, lacking real-time dialog with customers. Using dialog as an interactive method can help capture user preferences more accurately and enhance system transparency. However, developing such a goal-oriented dialog system has suffered many challenges as the system must collaborate with other subtasks, such as collecting user demands through interaction and recommending appropriate products to users. Additionally, most previous studies on dialog systems do not consider this situation and its challenges. This paper proposes a novel memory network framework for conversational recommendation, which harnesses dialog historical information to endow our model with adaptability in various dialog scenarios. Additionally, it leverages the knowledge base and user profiles to reweight candidates, reducing the ambiguity during interactions and improving the quality of conversational recommender systems. We demonstrate that the proposed method can achieve state-of-the-art performance in a few traditional tasks, such as options display and information provision, through experiments on the personalized bAbI dialog dataset and restaurant recommendation application. | IN SCOPE |
Modification of Neural network architecture / The selection mechanism of personalized response generation | Knowledge and Information Systems | 2023 | Personalized Dialogue Generation |
| Enhancing User Personalization in Conversational Recommenders | Conversational recommenders are emerging as a powerful tool to personalize a user's recommendation experience. Through a back-and-forth dialogue, users can quickly hone in on just the right items. Many approaches to conversational recommendation, however, only partially explore the user preference space and make limiting assumptions about how user feedback can be best incorporated, resulting in long dialogues and poor recommendation performance. In this paper, we propose a novel conversational recommendation framework with two unique features: (i) a greedy NDCG attribute selector, to enhance user personalization in the interactive preference elicitation process by prioritizing attributes that most effectively represent the actual preference space of the user; and (ii) a user representation refiner, to effectively fuse together the user preferences collected from the interactive elicitation process to obtain a more personalized understanding of the user. Through extensive experiments on four frequently used datasets, we find the proposed framework not only outperforms all the state-of-the-art conversational recommenders (in terms of both recommendation performance and conversation efficiency), but also provides a more personalized experience for the user under the proposed multi-groundtruth multi-round conversational recommendation setting. | IN SCOPE | The selection mechanism of personalized response generation / Modification of Neural network architecture / Personalization of adaptive circumstance/context
| WWW | 2023 | Personalized Dialogue Generation |
| Character adaptation of spoken dialogue systems based on user personalities | Character expression (e.g., extrovert or agreeable) is important for spo-ken dialogue systems to achieve human-like dialogue. The appropriate character isdifferent depending on each dialogue task and the user. In this study, we proposea character expression method according to the user personality in task-orienteddialogues. A previous psychological study identified four representative characterclasses based on the large-scale ratings on the Big Five traits. We use these four-character classes for character adaptation to the user personalities. Specifically, weinvestigate how the combination of the user personality and the system character af-fects the impression of the dialogue. Our analysis of a human-robot dialogue corpususing the Wizard of Oz (WOZ) method shows that the combination of the sub-ject personality and the robot character affects the favorable impressions toward therobot. Based on the analysis, we have designed and developed a character adapta-tion model that controls spoken dialogue behaviors: utterance amount, backchannelfrequency, fillers frequency and switching pause length. In a subjective experiment,a robot talked with subjects as a laboratory guide in four different character condi-tions, and each subject evaluated the impression of each robot. The results showsthat the extrovert character was preferred for items on the laboratory guide’s skill,and that the appropriate character to the user personality was preferred for items onhow easy to talk with the robot | IN SCOPE |
Personas for the agent | IWSDS | 2023 | Personalized Dialogue Generation |
| Topic-Enhanced Personalized Retrieval-Based Chatbot | Building a personalized chatbot has drawn much attention recently. A personalized chatbot is considered to have a consistent personality. There are two types of methods to learn the personality. The first mainly model the personality from explicit user profiles (e.g., manually created persona descriptions). The second learn implicit user profiles from the user’s dialogue history, which contains rich, personalized information. However, a user’s dialogue history can be long and noisy as it contains long-time, multi-topic historical dialogue records. Such data noise and redundancy impede the model’s ability to thoroughly and faithfully learn a consistent personality, especially when applied with models that have an input length limit (e.g., BERT). In this paper, we propose deconstructing the long and noisy dialogue history into topic-dependent segments. We only use the topically related dialogue segment as context to learn the topic-aware user personality. Specifically, we design a Topic-enhanced personalized Retrieval-based Chatbot, TopReC. It first deconstructs the dialogue history into topic-dependent dialogue segments and filters out irrelevant segments to the current query via a Heter-Merge-Reduce framework. It then measures the matching degree between the response candidates and the current query conditioned on each topic-dependent segment. We consider the matching degree between the response candidate and the cross-topic user personality. The final matching score is obtained by combining the topic-dependent and cross-topic matching scores. Experimental results on two large dataset show that TopReC outperforms all previous state-of-the-art methods. | IN SCOPE | | ECIR | 2023 | personalized chatbot |