Complex Factoid Question Answering with a Free-Text Knowledge Graph

Abstract

We introduce delft, a factoid question answering system which combines the nuance and depth of knowledge graph question answering approaches with the broader coverage of free-text. delft builds a free-text knowledge graph from Wikipedia, with entities as nodes and sentences in which entities co-occur as edges. For each question, delft finds the subgraph linking question entity nodes to candidates using text sentences as edges, creating a dense and high coverage semantic graph. A novel graph neural network reasons over the free-text graph-combining evidence on the nodes via information along edge sentences-to select a final answer. Experiments on three question answering datasets show delft can answer entity-rich questions better than machine reading based models, bert-based answer ranking and memory networks. delft's advantage comes from both the high coverage of its free-text knowledge graph-more than double that of dbpedia relations-and the novel graph neural network which reasons on the rich but noisy free-text evidence.

Dynamic Semantic Graph Construction and Reasoning for Explainable Multi-hop Science Question Answering

“…On the other hand, KG is able to provide structural clues about relevant entities for explainable predictions (Feng et al, 2020;Saxena et al, 2020;. But it is known to suffer from sparsity, where complex question clues are unlikely to be covered by the closed-form relations in KG (Zhao et al, 2020;Zhang et al, 2020b). Another issue is that KG requires large human labor and is easy to become outdated if not maintained timely.…”

“…In addition, previous works also explore the effectiveness of structured knowledge by either encoding the nodes (Yang and Mitchell, 2017;, triples (Mihaylov and Frank, 2018;), paths (Lin et al, 2019Lei et al, 2020) or tabular (Zhu et al, 2021a) (Saha and Mausam, 2018) is widely used in knowledge base question answering to extract entity-relation triples (Bosselut et al, 2019;Zhao et al, 2020;Deng et al, 2019). However, OpenIE favors precision over recall, which is not necessarily effective to form connections among diverse evidence facts for multi-hop QA.…”

Answering Open-Domain Questions of Varying Reasoning Steps from Text

“…Despite this success, most previous systems are developed with, and evaluated on, datasets that contain exclusively single-hop questions (ones that require a single document or paragraph to answer) or two-hop ones. As a result, their design is often tailored exclusively to single-hop (e.g., Chen et al, 2017;Wang et al, 2018b) or multi-hop questions (e.g., Nie et al, 2019;Min et al, 2019;Feldman and El-Yaniv, 2019;Zhao et al, 2020a;Xiong et al, 2021). Even when the model is designed to work with both, it is often trained and evaluated on exclusively single-hop or multi-hop settings (e.g., Asai et al, 2020).…”

“…Besides the impractical assumption about reasoning hops, previous work often also assumes access to non-textual metadata such as knowledge bases, entity linking, and Wikipedia hyperlinks when retrieving supporting facts, especially in answering complex questions (Nie et al, 2019;Feldman and El-Yaniv, 2019;Zhao et al, 2019;Asai et al, 2020;Dhingra et al, 2020;Zhao et al, 2020a). While this information is helpful, it is not always available in text collections we might be interested in getting answers from, such as news or academic research articles, besides being labor-intensive and time-consuming to collect and maintain.…”

Q. Can Knowledge Graphs be used to Answer Boolean Questions? A. It’s complicated!

“…(1) Question: is anne with an e filmed on pei Passage: The series is filmed partially in Prince Edward Island as well as ... Gold Answer: Yes Predicted Answer: No A number of works including Mihaylov and Frank (2018); Bauer et al (2018); Lin et al (2019); Qiu et al (2019); Thayaparan et al (2019); Talmor et al (2019); Zhao et al (2020) show successful usage of knowledge graphs (KGs) in several MRC settings. We propose and evaluate two approaches for augmenting questions and answers with KG information: (1) concatenating the model input with sentences constructed from ConceptNet triples 1 (Speer et al, 2017); and (2) encoding KG entities and relations with the Graph Neural Network (GNN) proposed by Shaw et al (2019), a model suited to graph-based input.…”