IEEE VIS 2024 Content: LinkQ: An LLM-Assisted Visual Interface for Knowledge Graph Question-Answering

LinkQ: An LLM-Assisted Visual Interface for Knowledge Graph Question-Answering

Harry Li - MIT Lincoln Laboratory, Lexington, United States

Gabriel Appleby - Tufts University, Medford, United States

Ashley Suh - MIT Lincoln Laboratory, Lexington, United States

Screen-reader Accessible PDF
Download preprint PDF
Download camera-ready PDF
Download Supplemental Material

Room: Bayshore VI

2024-10-17T18:30:00ZGMT-0600Change your timezone on the schedule page
2024-10-17T18:30:00Z
Exemplar figure, described by caption below
Exemplar workflow for LinkQ, a system leveraging an LLM for refining natural language questions into knowledge graph queries. The (A) Chat Panel lets users communicate with the LLM to ask specific or open-ended questions. The Query Preview Panel consists of three components: the (B1) Query Editor, which supports interactive editing; the (B2) Entity-Relation Table, which provides mapped data IDs from the KG, helping to assess the correctness of the LLM's generated query; and the (B3) Query Graph, which visualizes the structure of the query to illustrate the underlying schema of the KG. Finally, the (C) Results Panel provides a cleaned, exportable table as well as an LLM-generated summary based on the query results. Importantly, LinkQ ensures all data retrieved and summarized by the LLM comes from ground truth in the KG.
Fast forward
Keywords

Knowledge graphs, large language models, query construction, question-answering, natural language interfaces.

Abstract

We present LinkQ, a system that leverages a large language model (LLM) to facilitate knowledge graph (KG) query construction through natural language question-answering. Traditional approaches often require detailed knowledge of a graph querying language, limiting the ability for users - even experts - to acquire valuable insights from KGs. LinkQ simplifies this process by implementing a multistep protocol in which the LLM interprets a user's question, then systematically converts it into a well-formed query. LinkQ helps users iteratively refine any open-ended questions into precise ones, supporting both targeted and exploratory analysis. Further, LinkQ guards against the LLM hallucinating outputs by ensuring users' questions are only ever answered from ground truth KG data. We demonstrate the efficacy of LinkQ through a qualitative study with five KG practitioners. Our results indicate that practitioners find LinkQ effective for KG question-answering, and desire future LLM-assisted exploratory data analysis systems.