Knowledge graphs (KGs) provide a structured representation of real-world facts as triples consisting of entities and their relationships.With the rapid progress of large language models (LLMs), recent studies increasingly explore LLMs for end-to-end KG construction from text.In particular, generative knowledge extraction (GKE) builds KGs by directly generating structured triples from documents.However, generation errors are inevitable, and the resulting KGs often contain triples that do not align with the facts expressed in the source text.To address these issues, we propose GraphRefine, a framework that performs triple-level refinement on KGs constructed via GKE.We first analyze factual inconsistencies that arise in GKE and categorize their types based on a human evaluation.We then construct training data reflecting these types and fine-tune an LLM as a KG refiner.Given a draft KG, the fine-tuned refiner selects a refinement operation for each triple and, if needed, deletes, edits, or rewrites it to reduce factual inconsistencies.Extensive experiments demonstrate that GraphRefine goes beyond deletion-only approaches and improves KG quality from diverse perspectives.Table 8: Qualitative comparison of knowledge extraction results (Draft KG vs. GraphJudge vs. GraphRefine).

Unlike the traditional recommender systems that rely on historical data such as clicks or purchases, a conversational recommender system (CRS) aims to provide a personalized recommendation through a natural conversation. The conversational interaction facilitates capturing not only explicit preference from mentioned items but also implicit states, such as a user’s current situation and emotional states from a dialogue context. Nevertheless, existing CRSs fall short of fully exploiting a dialogue context since they primarily derive explicit user preferences from the items and item-attributes mentioned in a conversation. To address this limitation and attain a comprehensive understanding of a dialogue context, we proposeCoreSense, aconversationalrecommender system enhanced with social commonsenseknowledge. In other words, CoreSense exploits the social commonsense knowledge graph ATOMIC to capture the user’s implicit states, such as a user’s current situation and emotional states, from a dialogue context. Thus, the social commonsense knowledge-augmented CRS can provide a more appropriate recommendation from a given dialogue context. Furthermore, we enhance the collaborative filtering effect by utilizing the user’s states inferred from commonsense knowledge as an improved criterion for retrieving other dialogues of similar interests. Extensive experiments on CRS benchmark datasets show that CoreSense provides human-like recommendations and responses based on inferred user states, achieving significant performance improvements.

Numerous large-scale knowledge graphs (KGs) fundamentally represent two-view KGs: an ontology-view KG with abstract classes in ontology and an instance-view KG with specific collections of entities instantiated from ontology classes. Two-view KG embedding aims to jointly learn continuous vector representations of entities and relations in the aforementioned two-view KGs. In essence, an ontology schema exhibits a tree-like structure guided by class hierarchies, which leads classes to form inheritance hierarchies. However, existing two-view KG embedding models neglect those hierarchies, which provides the necessity to reflect class inheritance. On the other hand, KG is constructed based on a pre-defined ontology schema that includes heterogeneous relations between classes. Furthermore, these relations are defined within the scope of those among classes since instances inherit all the properties of their corresponding classes, which reveals structural similarity between two multi-relational networks. Despite the consideration to bridge the gap among two-view KG representations, existing methods ignore the existence of structural similarity between two-view KGs. To address these issues, we propose a novel two-view KG embedding model, CISS, considering Class Inheritance and Structural Similarity between two-view KGs. To deal with class inheritance, we utilize class sets, each of which is composed of sibling classes, to learn fine-grained class representations. In addition, we configure virtual instance-view KG from clustered instances and compare subgraph representations of two-view KGs to enhance structural similarity between them. Experimental results show our superior performance compared to existing models.

Conversational recommender systems (CRSs) capture a user preference through a conversation.However, the existing CRSs lack capturing comprehensive user preferences.This is because the items mentioned in a conversation are mainly regarded as a user preference.Thus, they have limitations in identifying a user preference from a dialogue context expressed without preferred items.Inspired by the characteristic of an online recommendation community where participants identify a context of a recommendation request and then comment with appropriate items, we exploit the Reddit data.Specifically, we propose a Contrastive Learning approach for Injecting Contextual Knowledge (CLICK) from the Reddit data to the CRS task, which facilitates the capture of a context-level user preference from a dialogue context, regardless of the existence of preferred item-entities.Moreover, we devise a relevance-enhanced contrastive learning loss to consider the fine-grained reflection of multiple recommendable items.We further develop a response generation module to generate a persuasive rationale for a recommendation.Extensive experiments on the benchmark CRS dataset show the effectiveness of CLICK, achieving significant improvements over stateof-the-art methods.

Conversational recommender systems (CRS) aim to acquire a user’s dynamic interests for a successful recommendation. By asking about his/her preferences, CRS explore current needs of a user and recommend items of interest. However, previous works may not determine a proper action in a timely manner which leads to the insufficient information gathering and the waste of conversation turns. Since they learn a single decision policy, it is difficult for them to address the general decision problems in CRS. Besides, existing methods do not distinguish whether the past behaviors inferred from the historical interactions are closely related to the user’s current preference. To address these issues, we propose a novel Hierarchical policy learning based Conversational Recommendation framework (HiCR). HiCR formulates the multi-round decision making process as a hierarchical policy learning scheme, which consists of both a high-level policy and a low-level policy. In detail, the high-level policy aims to determine what type of action to take, such as a recommendation or a query, by observing the comprehensive conversation information. According to the decided action type, the low-level policy selects a specific action, such as which attribute to ask or which item to recommend. The hierarchical conversation policy enables CRS to decide an optimal action, resulting in reducing the unnecessary consumption of conversation turns and the continuous failure of recommendations. Furthermore, in order to filter out the unnecessary historical information when enriching the current user preference, we extract and utilize the informative past behaviors that are attentive to the current needs. Empirical experiments on four real-world datasets show the superiority of our approach against the current state-of-the-art methods.