Conversational Recommender Systems Research Articles

Knowledge Graphs and Pretrained Language Models Enhanced Representation Learning for Conversational Recommender Systems.

Conversational recommender systems (CRSs) utilize natural language interactions and dialog history to infer user preferences and provide accurate recommendations. Due to the limited conversation context and background knowledge, existing CRSs rely on external sources such as knowledge graphs (KGs) to enrich the context and model entities based on their interrelations. However, these methods ignore the rich intrinsic information within entities. To address this, we introduce the knowledge-enhanced entity representation learning (KERL) framework, which leverages both the KG and a pretrained language model (PLM) to improve the semantic understanding of entities for CRS. In our KERL framework, entity textual descriptions are encoded via a PLM, while a KG helps reinforce the representation of these entities. We also employ positional encoding to effectively capture the temporal information of entities in a conversation. The enhanced entity representation is then used to develop a recommender component that fuses both entity and contextual representations for more informed recommendations, as well as a dialog component that generates informative entity-related information in the response text. A high-quality KG with aligned entity descriptions is constructed to facilitate this study, namely, the Wiki Movie Knowledge Graph (WikiMKG). The experimental results show that KERL achieves state-of-the-art results in both recommendation and response generation tasks. Our code is publicly available at the link: https://github.com/icedpanda/KERL.

Multi-source information contrastive learning collaborative augmented conversational recommender systems

AbstractConversational Recommender Systems (CRS) aim to provide high-quality items to users in fewer conversation rounds using natural language. Despite various attempts that have been made, there are still some problems: Previous CRS only learned item representations in a single knowledge graph and ignored item tags; information gaps exist in the same items from different knowledge graphs and information popularity both affect user preferences; system generated responses lack descriptiveness and diversity. To address these problems and fully utilize external knowledge, we propose a Multi-source Information Contrastive Learning Collaborative Augmented method ($$\textbf{MCCA}$$ MCCA ), which aims to mine the potential tag preferences of users in dialogues as well as enhance the accuracy of item representation and user preference modeling. Specifically, we utilize the obtained items and their tags to construct a new knowledge graph that incorporates movie tags. We design a Multi-source Item Fusion mechanism ($$\textbf{MIF}$$ MIF ) to bridge the information gaps between items from different knowledge graphs and then utilize unsupervised contrastive learning to enhance the items’ representation capability after MIF. Additionally, a Multi-Tag Fusion mechanism ($$\textbf{MTF}$$ MTF ) is designed to combine user-perceived information (i.e., tag popularity) and keywords obtained from reviews to co-enhance user preference representations through items and tags, and to incorporate fused item and tag features into the conversation module. Extensive experiments on two datasets show that MCCA significantly outperforms state-of-the-art methods. The source code will be available at https://github.com/lhy-cqut/MCCA.

Graph-based dynamic attribute clipping for conversational recommendation

Conversational recommender systems (CRS) enable traditional recommender systems to interact with users by asking questions about their preferences and recommending items. Conversation recommendation has made significant progress, however, studies on attribute-aware conversational recommendation have overlooked the problem that after obtaining the candidate attribute set, the relevance of the attributes in the candidate attribute set to the user’s current preferred attributes is not further considered, resulting in the existence of many user-uninterested attributes in the candidate attribute set. This seriously affects the dialogue quality and reduces the user’s patience, decreasing recommendation accuracy. To address this problem, this paper proposes a new framework called Attribute Clipping based on Dynamic Graph (ACDG). In ACDG, firstly, the reasoning module uses the restriction property of graph structure to obtain a large set of candidate attributes, and then the attribute clipping module filters out the attributes with high relevance to the current user’s preferred attributes. Thus, ACDG can obtain a better-quality set of candidate attributes. Through extensive experiments on four benchmark CRS datasets, we validate the effectiveness of the method.

Generating Usage-related Questions for Preference Elicitation in Conversational Recommender Systems

A key distinguishing feature of conversational recommender systems over traditional recommender systems is theirability to elicit user preferences using natural language. Currently, the predominant approach to preference elicitation is to ask questions directly about items or item attributes. Users searching for recommendations may not have deep knowledge of the available options in a given domain. As such, they might not be aware of key attributes or desirable values for them. However, in many settings, talking about the planned use of items does not present any difficulties, even for those that are new to a domain. In this article, we propose a novel approach to preference elicitation by asking implicit questions based on item usage. As one of the main contributions of this work, we develop a multi-stage data annotation protocol using crowdsourcing, to create a high-quality labeled training dataset. Another main contribution is the development of four models for the question generation task: two template-based baseline models and two neural text-to-text models. The template-based models use heuristically extracted common patterns found in the training data, while the neural models use the training data to learn to generate questions automatically. Using common metrics from machine translation for automatic evaluation, we show that our approaches are effective in generating elicitation questions, even with limited training data. We further employ human evaluation for comparing the generated questions using both pointwise and pairwise evaluation designs. We find that the human evaluation results are consistent with the automatic ones, allowing us to draw conclusions about the quality of the generated questions with certainty. Finally, we provide a detailed analysis of cases where the models show their limitations.

Laptop Recommender System Using the Hybrid of Ontology-Based and Collaborative Filtering

In the era of ever-evolving information technology, choosing the best laptop can be a complicated task for many users. The increasing complexity of technical specifications is often an obstacle, especially for users who need help understanding them. In addressing this challenge, we propose a solution: a laptop recommendation system that considers users' preferences and functional needs. We designed this system to help users choose a laptop that suits their daily functional needs. This system uses a form of Conversational Recommender System (CRS) by combining Ontology-Based Recommender System Filtering and Collaborative Filtering (CF). Ontology-Based Recommender System Filtering ensures a strong relationship between functional needs and technical specifications of laptops, making it easier for users to identify the right laptop. At the same time, Collaborative Filtering (CF) can provide diversity to the recommended products by using similar user preference data. We evaluate the accuracy of our system by calculating the success rate of recommendation accuracy with the accuracy metric, and the evaluation results show that the success rate of recommendation accuracy reaches 93.33%. Our system is highly effective in assisting users in choosing a laptop that suits their functional needs. With our laptop recommendation system, users can confidently select the correct laptop without being burdened by technical specifications, thus making their lives easier and more productive.

FacetCRS: Multi-Faceted Preference Learning for Pricking Filter Bubbles in Conversational Recommender System

The filter bubble is a notorious issue in Recommender Systems (RSs), which describes the phenomenon whereby users are exposed to a limited and narrow range of information or content that reinforces their existing dominant preferences and beliefs. This results in a lack of exposure to diverse and varied content. Many existing works have predominantly examined filter bubbles in static or relatively-static recommendation settings. However, filter bubbles will be continuously intensified over time due to the feedback loop between the user and the system in the real-world online recommendation. To address these issues, we propose a novel paradigm, Multi-Facet Preference Learning for Pricking Filter Bubbles in Conversational Recommender System (FacetCRS), which aims to burst filter bubbles in the conversational recommender system (CRS) through timely user-item interactions via natural language conversations. By considering diverse user preferences and intentions, FacetCRS automatically model user preference into multi-facets, including entity-, word-, context-, and review-facet, to capture diverse and dynamic user preferences to prick filter bubbles in the CRS. It is an end-to-end CRS framework to adaptively learn representations of various levels of preference facet and diverse types of external knowledge. Extensive experiments on two publicly available benchmark datasets demonstrate that our proposed method achieves state-of-the-art performance in mitigating filter bubbles and enhancing recommendation quality in CRS.

RecWizard: A Toolkit for Conversational Recommendation with Modular, Portable Models and Interactive User Interface

We present a new Python toolkit called RecWizard for Conversational Recommender Systems (CRS). RecWizard offers support for development of models and interactive user interface, drawing from the best practices of the Huggingface ecosystems. CRS with RecWizard are modular, portable, interactive and Large Language Models (LLMs)-friendly, to streamline the learning process and reduce the additional effort for CRS research. For more comprehensive information about RecWizard, please check our GitHub https://github.com/McAuley-Lab/RecWizard.

A Goal Interaction Graph Planning Framework for Conversational Recommendation

Multi-goal conversational recommender system (MG-CRS) which is more in line with realistic scenarios has attracted a lot of attention. MG-CRS can dynamically capture the demands of users in conversation, continuously engage their interests, and make recommendations. The key of accomplishing these tasks is to plan a reasonable goal sequence which can naturally guide the user to accept the recommended goal. Previous works have demonstrated that mining the correlations of goals from the goal sequences in the dialogue corpus is helpful for recommending the goal that the user is interested in. However, they independently model correlations for each level of goal (i.e., goal type or entity) and neglect the order of goals appear in the dialogue. In this paper, we propose a goal interaction graph planning framework which constructs a directed heterogeneous graph to flexibly model the correlations between any level of goals and retain the order of goals. We design a goal interaction graph learning module to model the goal correlations and propagate goal representations via directed edges, then use an encoder and a dual-way fusion decoder to extract the most relevant information with the current goal from the conversation and domain knowledge, making the next-goal prediction fully exploit the prior goal correlations and user feedback. Finally we generate engaging responses based on the predicted goal sequence to complete the recommendation task. Experiments on two benchmark datasets show that our method achieves significant improvements in both the goal planning and response generation tasks.

Conversational recommender based on graph sparsification and multi-hop attention

Conversational recommender systems provide users with item recommendations via interactive dialogues. Existing methods using graph neural networks have been proven to be an adequate representation of the learning framework for knowledge graphs. However, the knowledge graph involved in the dialogue context is vast and noisy, especially the noise graph nodes, which restrict the primary node’s aggregation to neighbor nodes. In addition, although the recurrent neural network can encode the local structure of word sequences in a dialogue context, it may still be challenging to remember long-term dependencies. To tackle these problems, we propose a sparse multi-hop conversational recommender model named SMCR, which accurately identifies important edges through matching items, thus reducing the computational complexity of sparse graphs. Specifically, we design a multi-hop attention network to encode dialogue context, which can quickly encode the long dialogue sequences to capture the long-term dependencies. Furthermore, we utilize a variational auto-encoder to learn topic information for capturing syntactic dependencies. Extensive experiments on the travel dialogue dataset show significant improvements in our proposed model over the state-of-the-art methods in evaluating recommendation and dialogue generation.

Dual-space Hierarchical Learning for Goal-guided Conversational Recommendation

Proactively and naturally guiding the dialog from the non-recommendation context (e.g., Chit-chat) to the recommendation scenario (e.g., Music) is crucial for the Conversational Recommender System (CRS). Prior studies mainly focus on planning the next dialog goal (e.g., chat on a movie star) conditioned on the previous dialog. However, we find the dialog goals can be simultaneously observed at different levels, which can be utilized to improve CRS. In this paper, we propose Dual-space Hierarchical Learning (DHL) to leverage multi-level goal sequences and their hierarchical relationships for conversational recommendation. Specifically, we exploit multi-level goal sequences from both the representation space and the optimization space. In the representation space, we propose the hierarchical representation learning where a cross attention module derives mutually enhanced multi-level goal representations. In the optimization space, we devise the hierarchical weight learning to reweight lower-level goal sequences, and introduce bi-level optimization for stable update. Additionally, we propose a soft labeling strategy to guide optimization gradually. Experiments on two real-world datasets verify the effectiveness of our approach. Code and data are available here.

Improving conversational recommender systems via multi-preference modelling and knowledge-enhanced

Conversational recommender systems (CRS) aim to capture user preferences and provide high-quality recommendations through conversations. Existing CRSs use items in conversations to represent user preferences, and these items usually need to match corresponding entities in external knowledge graphs (KGs). However, current general approaches extract subgraphs from large KGs (e.g., DBpedia) and thus may contain noise from nonrelevant domains and do not cover items in the CRS domain well. In addition, existing CRSs do not focus on hidden semantic preferences in user conversations, nor do they emphasize item attributes. Moreover, the system-generated responses lack diversity and descriptiveness. To address these issues, we propose a novel approach for Improving Conversation Recommendation Systems via Multi-Preference Modelling and Knowledge-Enhanced (MPKE). Specifically, we first customize a higher-quality knowledge graph LMKG for CRS to help the model fully exploit external knowledge to improve its performance. Second, we design a semantic fusion module to augment the semantic connections between external knowledge and dialogue text. Then, we propose a two-stage recommendation mechanism to explore more item attributes and fuse entity and text semantics to enhance user representation. For response generation, we use a word-level KG (i.e., ConceptNet) to enhance the representation of conversation keywords and construct explanation templates with slots to enhance the diversity of system responses. extensive experiments show that our approach outperforms past state-of-the-art methods. the source code will be published at https://github.com/zcy-cqut/MPKE.

Improving Conversational Recommender System via Contextual and Time-Aware Modeling with Less Domain-Specific Knowledge

Improving Conversational Recommender System via Contextual and Time-Aware Modeling with Less Domain-Specific Knowledge

UCRI: A Unified Conversational Recommender System Based on Item-Guided Conditional Generation

UCRI: A Unified Conversational Recommender System Based on Item-Guided Conditional Generation

Effects of User-Reported Risk Factors and Follow-Up Care Activities on Satisfaction With a COVID-19 Chatbot: Cross-Sectional Study.

The COVID-19 pandemic influenced many to consider methods to reduce human contact and ease the burden placed on health care workers. Conversational agents or chatbots are a set of technologies that may aid with these challenges. They may provide useful interactions for users, potentially reducing the health care worker burden while increasing user satisfaction. Research aims to understand these potential impacts of chatbots and conversational recommender systems and their associated design features. The objective of this study was to evaluate user perceptions of the helpfulness of an artificial intelligence chatbot that was offered free to the public in response to COVID-19. The chatbot engaged patients and provided educational information and the opportunity to report symptoms, understand personal risks, and receive referrals for care. A cross-sectional study design was used to analyze 82,222 chats collected from patients in South Carolina seeking services from the Prisma Health system. Chi-square tests and multinomial logistic regression analyses were conducted to assess the relationship between reported risk factors and perceived chat helpfulness using chats started between April 24, 2020, and April 21, 2022. A total of 82,222 chat series were started with at least one question or response on record; 53,805 symptom checker questions with at least one COVID-19-related activity series were completed, with 5191 individuals clicking further to receive a virtual video visit and 2215 clicking further to make an appointment with a local physician. Patients who were aged >65 years (P<.001), reported comorbidities (P<.001), had been in contact with a person with COVID-19 in the last 14 days (P<.001), and responded to symptom checker questions that placed them at a higher risk of COVID-19 (P<.001) were 1.8 times more likely to report the chat as helpful than those who reported lower risk factors. Users who engaged with the chatbot to conduct a series of activities were more likely to find the chat helpful (P<.001), including seeking COVID-19 information (3.97-4.07 times), in-person appointments (2.46-1.99 times), telehealth appointments with a nearby provider (2.48-1.9 times), or vaccination (2.9-3.85 times) compared with those who did not perform any of these activities. Chatbots that are designed to target high-risk user groups and provide relevant actionable items may be perceived as a helpful approach to early contact with the health system for assessing communicable disease symptoms and follow-up care options at home before virtual or in-person contact with health care providers. The results identified and validated significant design factors for conversational recommender systems, including triangulating a high-risk target user population and providing relevant actionable items for users to choose from as part of user engagement.

Explicit Knowledge Graph Reasoning for Conversational Recommendation

Traditional recommender systems estimate user preference on items purely based on historical interaction records, thus failing to capture fine-grained yet dynamic user interests and letting users receive recommendation only passively. Recent conversational recommender systems (CRSs) tackle those limitations by enabling recommender systems to interact with the user to obtain her/his current preference through a sequence of clarifying questions. Recently, there has been a rise of using knowledge graphs (KGs) for CRSs, where the core motivation is to incorporate the abundant side information carried by a KG into both the recommendation and conversation processes. However, existing KG-based CRSs are subject to two defects: (1) there is a semantic gap between the learned representations of utterances and KG entities, hindering the retrieval of relevant KG information; (2) the reasoning over KG is mostly performed with the implicitly learned user interests, overlooking the explicit signals from the entities actually mentioned in the conversation. To address these drawbacks, we propose a new CRS framework, namely the K nowledge E nhanced C onversational R easoning (KECR) model. As a user can reflect her/his preferences via both attribute- and item-level expressions, KECR jointly embeds the structured knowledge from two levels in the KG. A mutual information maximization constraint is further proposed for semantic alignment between the embedding spaces of utterances and KG entities. Meanwhile, KECR utilizes the connectivity within the KG to conduct explicit reasoning of the user demand, making the model less dependent on the user’s feedback to clarifying questions. As such, the semantic alignment and explicit KG reasoning can jointly facilitate accurate recommendation and quality dialogue generation. By comparing with strong baselines on two real-world datasets, we demonstrate that KECR obtains state-of-the-art recommendation effectiveness, as well as competitive dialogue generation performance.

Understanding and Predicting User Satisfaction with Conversational Recommender Systems

User satisfaction depicts the effectiveness of a system from the user’s perspective. Understanding and predicting user satisfaction is vital for the design of user-oriented evaluation methods forconversational recommender systems (CRSs). Current approaches rely on turn-level satisfaction ratings to predict a user’s overall satisfaction with CRS. These methods assume that all users perceive satisfaction similarly, failing to capture the broader dialogue aspects that influence overall user satisfaction.We investigate the effect of several dialogue aspects on user satisfaction when interacting with a CRS. To this end, we annotate dialogues based on six aspects (i.e.,relevance,interestingness,understanding,task-completion,interest-arousal, andefficiency) at the turn and dialogue levels. We find that the concept of satisfaction varies per user. At the turn level, a system’s ability to make relevant recommendations is a significant factor in satisfaction. We adopt these aspects as features for predicting response quality and user satisfaction. We achieve an F1-score of 0.80 in classifying dissatisfactory dialogues, and a Pearson’srof 0.73 for turn-level response quality estimation, demonstrating the effectiveness of the proposed dialogue aspects in predicting user satisfaction and being able to identify dialogues where the system is failing.With this article, we release our annotated data.1

Enhancing conversational recommender systems via multi-level knowledge modeling with semantic relations

Augmenting conversational recommendation systems (CRS) with prior knowledge is crucial for learning user preferences and understanding contextual semantics. Existing methods are mainly based on knowledge graphs (KG), which rely on relationship information between entities, ignoring the semantic relationship in the dialogue. In addition, due to the limited information reflected by structured knowledge in KG, they still need broader external knowledge support to understand contextual semantics.To address the above issues, we propose a knowledge prompt-based CRS named MKCRS, which utilizes semantic relations to model multi-level knowledge and integrate it into CRS sub-tasks. Specifically, we first derive multi-level knowledge representations related to the current dialogue from external knowledge and then perform deep explicit and implicit modeling to construct knowledge prompts. To implement this framework, we design a semantic fusion model based on named entity phrases and a unified knowledge prompt generator. The former exploits explicit semantic relations between entities and contexts to enhance knowledge representations, while the latter adaptively learns the most important knowledge prompts via implicit connections between knowledge. Finally, based on the generated knowledge prompts, we construct knowledge-enhanced prompt templates for recommendation and dialogue generation tasks. Experimental results on the two datasets demonstrate that MKCRS outperforms the state-of-the-art models.

CRS-Que : A User-Centric Evaluation Framework for Conversational Recommender Systems

An increasing number of recommendation systems try to enhance the overall user experience by incorporating conversational interaction. However, evaluating conversational recommender systems (CRSs) from the user’s perspective remains elusive. The GUI-based system evaluation criteria may be inadequate for their conversational counterparts. This paper presents our proposed unifying framework, CRS-Que , to evaluate the user experience of CRSs. This new evaluation framework is developed based on ResQue , a popular user-centric evaluation framework for recommender systems. Additionally, it includes user experience metrics of conversation (e.g., understanding, response quality, humanness) under two dimensions of ResQue (i.e., Perceived Qualities and User Beliefs). Following the psychometric modeling method, we validate our framework by evaluating two conversational recommender systems in different scenarios: music exploration and mobile phone purchase . The results of the two studies support the validity and reliability of the constructs in our framework and reveal how conversation constructs and recommendation constructs interact and influence the overall user experience of the CRS. We believe this framework could help researchers conduct standardized user-centric research for conversational recommender systems and provide practitioners with insights into designing and evaluating a CRS from users’ perspectives.

Leveraging Long Short-Term User Preference in Conversational Recommendation via Multi-agent Reinforcement Learning

Conversational recommender systems (CRS) endow traditional recommender systems with the capability of dynamically obtaining users' short-term preferences for items and attributes through interactive dialogues. There are three core challenges for CRS, including the intelligent decisions for what attributes to ask, which items to recommend, and when to ask or recommend, at each conversation turn. Previous methods mainly leverage reinforcement learning (RL) to learn conversational recommendation policies for solving one or two of these three decision-making problems in CRS with separated conversation and recommendation components. These approaches restrict the scalability and generality of CRS and fall short of preserving a stable training procedure. In the light of these challenges, we tackle these three decision-making problems in CRS as a unified policy learning task. In order to leverage different features that are important to each sub-problem and facilitate better unified policy learning in CRS, we propose two novel multi-agent RL-based frameworks, namely Independent and Hierarchical Multi-Agent UNIfied COnversational RecommeNders (IMA-UNICORN and HMA-UNICORN), respectively. In specific, two low-level agents enrich the state representations for attribute prediction and item recommendation, by combining the long-term user preference information from the historical interaction data and the short-term user preference information from the conversation history. A high-level meta agent is responsible for coordinating the low-level agents to adaptively make the final decision. Experimental results on four benchmark CRS datasets and a real-world E-Commerce application show that the proposed frameworks significantly outperform state-of-the-art methods. Extensive analyses further demonstrate the superior scalability of the MARL frameworks on the multi-round conversational recommendation.

Improved conversational recommender system based on dialog context

Abstract Conversational recommender system (CRS) needs to be seamlessly integrated between the two modules of recommendation and dialog, aiming to recommend high-quality items to users through multiple rounds of interactive dialogs. Items can typically refer to goods, movies, news, etc. Through this form of interactive dialog, users can express their preferences in real time, and the system can fully understand the user’s thoughts and recommend corresponding items. Although mainstream dialog recommendation systems have improved the performance to some extent, there are still some key issues, such as insufficient consideration of the entity’s order in the dialog, the different contributions of items in the dialog history, and the low diversity of generated responses. To address these shortcomings, we propose an improved dialog context model based on time-series features. Firstly, we augment the semantic representation of words and items using two external knowledge graphs and align the semantic space using mutual information maximization techniques. Secondly, we add a retrieval model to the dialog recommendation system to provide auxiliary information for generating replies. We then utilize a deep timing network to serialize the dialog content and more accurately learn the feature relationship between users and items for recommendation. In this paper, the dialog recommendation system is divided into two components, and different evaluation indicators are used to evaluate the performance of the dialog component and the recommendation component. Experimental results on widely used benchmarks show that the proposed method is effective.