Knowledge graph is a hot topic in the Web Conference 2020, with tasks of entity alignment, relation extraction, knowledge graph completion and refinement intensively studied. This paper list briefs those nice works.
Some interesting findings:
-
Many papers propose methods exploiting rich information such as multi-type entities and neighborhood information, structural information and correlation and user-item interaction;
-
Low resource settings are paid great attention, for example, relation extraction by Zhou et al., MetaNER and KGC with adversarial network;
-
Rule mining is also an interesting direction, with papers like NERO and KGist;
-
Graph neural networks and adversarial learning are definitely among the most popular topics, such as [He et al.] and [Cao et al.].
Last but not least, there are a paper on storage architecture, a resource article of event knowledge graph, and three downstream applications of recommendation, QA, and keyword search.
Ads.: Know more about knowledge graphs, please check out our recent survey entitled A Survey on Knowledge Graphs: Representation, Acquisition and Applications!
Collective Multi-type Entity Alignment Between Knowledge Graphs WWW 2020. Zhu et al. [Paper]
Motivation: the sparsity of different knowledge graphs and alignment of multi-type entities jointly aligns multiple types of entities, collectively leverages the neighborhood information and generalizes to unlabeled entity types
MetaNER: Named Entity Recognition with Meta-Learning WWW 2020. Li et al. [Paper]
Meta-learning approach for domain adaptation in NER
Novel Entity Discovery from Web Tables WWW 2020. Zhang et al. [Paper]
leverage the content in such tables to discover new entities, properties, and relationships
High Quality Candidate Generation and Sequential Graph Attention Network for Entity Linking WWW 2020. Fang et al. [Paper]
considering the differences between candidate entities, promote the quality of candidates; utilize the topical coherence among the referred entities, distinguish candidate entities, capture the relevance between the current mention and its subsequent entities
Dynamic Graph Convolutional Networks for Entity Linking WWW 2020. Wu et al. [Paper]
Motivation: structured graph for a set of entities depends on the contextual information of the given document and adaptively changes on different aggregation layers of GCN Dynamic GCN
NERO: A Neural Rule Grounding Framework for Label-Efficient Relation Extraction WWW 2020. Zhou et al.. [Paper]
neural soft rule matching
LOREM: Language-consistent Open Relation Extraction from Unstructured Text. WWW 2020. Harting et al. [Paper]
Language-consistent multi-lingual Open Relation Extraction Model
Beyond Triplets: Hyper-Relational Knowledge Graph Embedding for Link Prediction WWW 2020. Rosso et al. [Paper]
captures not only the primary structural information of the KG encoded in the triplets, but also the correlation between each triplet and its associated key-value pairs.
Generalizing Tensor Decomposition for N-ary Relational Knowledge Bases WWW 2020. Liu et al.. [Paper]
generalize tensor decomposition for n-ary relational KBs
Relation Adversarial Network for Low Resource Knowledge Graph Completion WWW 2020. Zhang et al. [Paper]
low resource relations + adversarial network Adversarial transfer learning: weighted relation adversarial network; learn relation-invariant features more transferable from source relations to target relations; relation-gated mechanism fully relaxes the shared label space assumptions
Mining Implicit Entity Preference from User-Item Interaction Data for Knowledge Graph Completion via Adversarial Learning WWW 2020. He et al. [Paper]
leverage rick user-item interaction data adversarial learning, collaborative learning, graph neural networks
Open Knowledge Enrichment for Long-tail Entities. WWW 2020. Cao et al.. [Paper] [Code]
OKELE, enrich long-tail entities from the open Web, that is search the Web to find a set of true facts of a given long-tail entity; three steps: property prediction, value extraction, fact verification open knowledge enrichment on long-tail entities; GNN and graph attention; fact verification model based on a PGM with conjugate priors;
What is Normal, What is Strange, and What is Missing in a Knowledge Graph: Unified Characterization via Inductive Summarization WWW 2020. Belth et al. [Paper]
KGist, Knowledge Graph Inductive SummarizaTion learns a summary of inductive rules that best compress the KG according to the Minimum Description Length principle—a formulation that we are the first to use in the context of KG rule mining.
Correcting Knowledge Base Assertions. WWW 2020. Chen et al.. [Paper]
correction framework combines lexical matching, semantic embedding, soft constraint mining and semantic consistency checking.
Expanding Taxonomies with Implicit Edge Semantics WWW 2020. Manzoor et al. [Paper]
taxonomies with heterogeneous edge semantics unobserved
Adaptive Low-level Storage of Very Large Knowledge Graphs. WWW 2020. Urbani et al.. [Paper]
a novel storage structure for very large KGs on centralized systems
ASER: A Large-scale Eventuality Knowledge Graph WWW 2020. Zhang et al. [Paper]
ASER (activities, states, events, and their relations), a large-scale eventuality knowledge graph extracted from more than 11-billion-token unstructured textual data.
Reinforced Negative Sampling over Knowledge Graph for Recommendation WWW 2020. Wang et al. [Paper]
a reinforcement learning agent to explore high-quality negatives Motivation: item knowledge graph for inferring informative and factual negative samples Knowledge Graph Policy Network (KGPolicy): reinforcement learning agent to explore high-quality negatives; matrix factorization
Complex Factoid Question Answering with a Free-Text Knowledge Graph WWW 2020. Zhao et al. [Paper]
Keyword Search over Knowledge Graphs via Static and Dynamic Hub Labelings WWW 2020. Shi et al. [Paper]