Loading Events

« All Events

  • This event has passed.

Ph.D. Thesis {Colloquium}: CDS: 31st May 2022 : “Leveraging Knowledge Graph Embeddings for Question Answering.”

31 May @ 10:00 AM -- 11:00 AM

Ph.D. Thesis Colloquium

Speaker                 : Mr. Apoorv Umang Saxena

S.R. Number         : 06-18-02-10-12-18-1-15505
Title                        : Leveraging Knowledge Graph Embeddings for Question Answering
Research Supervisor :  Prof. Partha Pratim Talukdar
Date & Time          : May 31, 2022 (Tuesday), 10:00 AM
Venue                      : Online

Knowledge graphs (KG) are multi-relational graphs consisting of entities as nodes and relations among them as typed edges. The goal of knowledge graph question answering (KGQA) is to answer natural language queries posed over the KG. Most KGQA systems use variations of semantic parsing to achieve this, where a natural language query is translated to a structured query (e.g., SQL) and then executed over the knowledge base. However, KGs are often incomplete with many missing links,​ posing challenges for such parse-then-execute systems, especially as the complexity of the query increases. Recent research on KGQA has attempted to handle KG sparsity using relevant external text — which isn’t always readily available — or graph neural network-based methods, which struggle with scale. 

In a separate line of research, KG embedding methods (KGEs) have been proposed to reduce KG sparsity by performing missing link prediction. These methods aim to learn low-dimension representations of entities and relations in the KG, which are then utilized by various scoring functions to predict the plausibility of unknown triples. Such KG embedding methods, even though highly relevant, have not been explored for KGQA so far. In this work, we fill this gap by proposing methods to leverage KGEs in the KGQA pipeline. We do this through the following contributions:

 Improving Multi-Hop KGQA using KG Embeddings 

We first tackle a subset of KGQA queries — multi-hop KGQA. We propose EmbedKGQA, a method which uses ComplEx embeddings and scoring function to answer these queries. We find that EmbedKGQA is particularly effective at KGQA over sparse KGs, while it also relaxes the requirement of answer selection from a pre-specified local neighborhood, an undesirable constraint imposed by GNN-based for this task. Experiments show that EmbedKGQA is superior to several GNN-based methods on incomplete KGs across a variety of dataset scales.

 Question Answering over Temporal Knowledge Graphs 

We then extend our method to temporal knowledge graphs (TKG), where each edge in the KG is accompanied by a time scope (i.e. start and end times). Here, instead of KGEs, we make use of temporal KGEs (TKGE) to enable the model to make use of these time annotations and perform temporal reasoning. We also propose a new dataset – CronQuestions – which is one of the largest publicly available temporal KGQA dataset with over 400k template-based temporal reasoning questions. Through extensive experiments we show the superiority of our method, CronKGQA, over several language-model baselines on the challenging task of temporal KGQA on CronQuestions.

Sequence-to-Sequence Knowledge Graph Completion and Question Answering 

So far, integrating KGE into the KGQA pipeline had required separate training of the KGE and KGQA modules. In this work, we show that an off-the-shelf encoder-decoder Transformer model can serve as a scalable and versatile KGE model obtaining state-of-the-art results for KG link prediction and incomplete KG question answering. We achieve this by posing KG link prediction as a sequence-to-sequence task and exchange the triple scoring approach taken by prior KGE methods with autoregressive decoding. Such a simple but powerful method re- duces the model size up to 98\% compared to conventional KGE models while keeping inference time tractable. It also allows us to answer a variety of KGQA queries, not being restricted by query type. 


31 May
10:00 AM -- 11:00 AM