Triple Trustworthiness Measurement for Knowledge Graph

Abstract

e Knowledge graph (KG) uses the triples to describe the facts in the real world. It has been widely used in intelligent analysis and applications. However, possible noises and con icts are inevitably introduced in the process of constructing. And the KG based tasks or applications assume that the knowledge in the KG is completely correct and inevitably bring about potential deviations. In this paper, we establish a knowledge graph triple trustworthiness measurement model that quantify their semantic correctness and the true degree of the facts expressed. e model is a crisscrossing neural network structure. It synthesizes the internal semantic information in the triples and the global inference information of the KG to achieve the trustworthiness measurement and fusion in the three levels of entity level, relationship level, and KG global level. We analyzed the validity of the model output con dence values, and conducted experiments in the real-world dataset FB15K (from Freebase) for the knowledge graph error detection task. e experimental results showed that compared with other models, our model achieved signi cant and consistent improvements.

Link Trustworthiness Evaluation over Multiple Heterogeneous Information Networks

“…is method exploits the structure information of the network. KGTtm [10] is a multi-level knowledge graph link trustworthiness evaluation method. It separates evaluation into nodes, edges, and graph levels, combining the internal semantic information of links and the global inference information of the KB.…”

“…For instance, R-GCN models relational data in the GCN framework. On the other hand, edge learning methods [3,[8][9][10] measure the acceptability of a link composed of an edge and two nodes corresponding to it. For example, TransE [9] considers whether the embedding nodes and edges of multi-relational data converge in low-dimensional vector spaces.…”

Fast Computation of Explanations for Inconsistency in Large-Scale Knowledge Graphs

“…Among the statistics-based methods, the closest to ours are [22,27], where faulty triples are identified by exploiting statistical distributions of KG relations and types. Another relevant work from this category [15] introduces a machine learning model to measure the trustworthiness of triples.…”

“…A common task in such settings is that of detecting incorrect facts in KGs. Although many works on this topic have focused on machine-learning techniques (e.g., [15,27]), methods that rely on symbolic reasoning (e.g., [25,32]) have shown the benefit for this task due to their accuracy and human-interpretability [26]. One prominent such method for detecting incorrect facts in KGs is to compute inconsistency explanations of the KG with respect to a manually crafted ontology [14], which is a set of axioms that capture the relevant domain of interest and constraints enforcing that some statements must not occur together.…”

Embedding Learning with Triple Trustiness on Noisy Knowledge Graph

“…Dong et al [ 11 ] propose a joint approach with both prior knowledge stemmed from KG and external web content to estimate triple quality in KG construction, but lacking flexible ability in scale and reasoning capability without embedding strategy. Jia et al [ 48 ] propose a crisscrossing neural network for KG completion and correction at the same time, while having high complexity and computational cost. In this paper, we introduce the triple trustiness for KGR, by considering the typical external heterogeneous source (i.e., entity type instances and entity descriptions) beyond the KG itself.…”