Encoding Temporal Information for Time-Aware Link Prediction

Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2350–2354, Austin, Texas, November 1-5, 2016. c⃝2016 Association for Computational Linguistics Encoding Temporal Information for Time-Aware Link Prediction Tingsong Jiang, Tianyu Liu, Tao Ge, Lei Sha, Sujian Li, Baobao Chang and Zhifang Sui Key Laboratory of Computational Linguistics, Ministry of Education School of Electronics Engineering and Computer Science, Peking University Collaborative Innovation Center for Language Ability, Xuzhou 221009 China {tingsong,tianyu0421,taoge,shalei,lisujian,chbb,szf}@pku.edu.cn Abstract Most existing knowledge base (KB) embed- ding methods solely learn from time-unknown fact triples but neglect the temporal informa- tion in the knowledge base. In this paper, we propose a novel time-aware KB embed- ding approach taking advantage of the hap- pening time of facts. Specifically, we use tem- poral order constraints to model transforma- tion between time-sensitive relations and en- force the embeddings to be temporally consis- tent and more accurate. We empirically eval- uate our approach in two tasks of link predic- tion and triple classification. Experimental re- sults show that our method outperforms other baselines on the two tasks consistently. 1 Introduction Knowledge bases (KBs) such as Freebase (Bollacker et al., 2008) and YAGO (Fabian et al., 2007) play a pivotal role in many NLP related applications. KB- s consist of facts in the form of triplets (ei, r, ej), indicating that head entity ei and tail entity ej is linked by relation r. Although KBs are large, they are far from complete. Link prediction is to predict relations between entities based on existing triplet- s, which can alleviate the incompleteness of cur- rent KBs. Recently a promising approach for this task called knowledge base embedding (Nickel et al., 2011; Bordes et al., 2011; Socher et al., 2013) aims to embed entities and relations into a continu- ous vector space while preserving certain informa- tion of the KB graph. TransE (Bordes et al., 2013) is a typical model considering relation vector as trans- lating operations between head and tail vector, i.e., ei + r ≈ ej when (ei, r, ej) holds. Most existing KB embedding methods solely learn from time-unknown facts but ignore the use- ful temporal information in the KB. In fact, there are many temporal facts (or events) in the KB, e.g., (Obama, wasBornIn, Hawaii) happened at August 4, 1961. (Obama, presidentOf, USA) is true since 2009. Current KBs such as YAGO and Freebase store such temporal information either directly or indirectly. The happening time of time-sensitive facts may indicate special temporal order of fact- s and time-sensitive relations. For example, (Ein- stein, wasBornIn, Ulm) happened in 1879, (Einstein, wonPrize, Nobel Prize) happened in 1922, (Einstein, diedIn, Princeton) occurred in 1955. We can infer the temporal order of time-sensitive relations from all such kinds of facts: wasBornIn → wonPrize → diedIn. Traditional KB embedding models such as TransE often confuse relations such as wasBornIn and diedIn when predicting (person,?,location) be- cause TransE learns only from time-unknown facts and cannot distinguish relations with similar seman- tic meaning. To make more accurate predictions, …

Content truncated to 500 words