Open Domain Question Answering System Based on Knowledge Base

Lai, Yongqing; Lin, Yang; Chen, Jiahao; Feng, Yansong; Zhao, Dongyan

doi:10.1007/978-3-319-50496-4_65

Cited by 21 publications

(11 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Experimental results show that our proposed method [17] achieves the third place among the 21 systems on NLPCC-ICCPOL 2016 KBQA challenge task [39]. The method proposed in [14] achieves the first place and the method proposed in [16] achieves the second place. The improved method proposed in this paper achieves a much better result than most of the state-of-the-art methods.…”

Section: Resultsmentioning

confidence: 90%

“…However, most of the information retrieval based methods use all possible n-grams of words of the question to retrieve the candidates from the knowledge base which can introduce lots of noise candidate triples. The system proposed by Lai et al [14] used a SPE (subject predicate extraction) algorithm to extract subject-predicate pair from the question and translate it to a KB query to search the candidates and use a method based on word vector similarity and predicate attention to score the candidate predicates. The method proposed by Wang et al [15] used a classifier to judge whether the predicate in the triple is what the question asked for.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Topic enhanced deep structured semantic models for knowledge base question answering

Xie

Zhao

Zhou

et al. 2017

Sci. China Inf. Sci.

View full text Add to dashboard Cite

Knowledge Base Question Answering (KBQA) is a hot research topic in natural language processing (NLP). The most challenging problem in KBQA is how to understand the semantic information of natural language questions and how to bridge the semantic gap between the natural language questions and the structured fact triples in knowledge base. This paper focuses on simple questions which can be answered by a single fact triple in knowledge base. We propose a topic enhanced deep structured semantic model for KBQA. The proposed method considers the task of KBQA as a matching problem between questions and the subjects and predicates in knowledge base. And the proposed model consists of two stages to match the subjects and predicates, respectively. In the first stage, we propose a Convolutional based Topic Entity Extraction Model (CTEEM) to extract topic entities mentioned in questions. With the extracted entities, we can retrieve the relevant candidate fact triples from knowledge base and obviously decrease the amount of noising candidates. In the second stage, we employ Deep Structured Semantic Models (DSSMs) to compute the semantic relevant score between questions and predicates in the candidates. And we combine the semantic level and the lexical level scores to rank the candidates. We evaluate the proposed method on KBQA dataset released by NLPCC-ICCPOL 2016. The experimental results show that our proposed method achieves the third place among the 21 submitted systems. Furthermore, we also extend the DSSM by using BiLSTM and integrate a convolutional structure on the top of BiLSTM layers. Our experimental results show that the extension models can further improve the performance.

show abstract

Section: Resultsmentioning

confidence: 90%

Section: Related Workmentioning

confidence: 99%

Topic enhanced deep structured semantic models for knowledge base question answering

Xie

Zhao

Zhou

et al. 2017

Sci. China Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…It can be found that our system outperforms all other methods. Compared with systems using pipeline model method and other sophisticated features and hand-craft rules (such as the use of part-of-speech features in the mention detection stage) [14], our framework can still improve the performance on KBQA task, and the straight-forward method is competitive with the state-of-thearts one. [43] 81.06 Kai Lei et al [17] 80.97 Lei Su et al [27] 79.41 Lai et al [14] 82.47 CTEEM+Combined-DSSM+LMS [34] 82.43 BB-KBQA [19] 84.12 MFSMM [42] 80.35 SiamsesATT [28] 81.81 CGRM 85.04…”

Section: Kbqamentioning

confidence: 99%

Calculating Question Similarity is Enough: A New Method for KBQA Tasks

Zhao¹,

Yuan²,

Jiahong³

et al. 2021

Preprint

View full text Add to dashboard Cite

Knowledge Base Question Answering (KBQA) aims to answer natural language questions with the help of an external knowledge base. The core idea is to find the link between the internal knowledge behind questions and known triples of the knowledge base. The KBQA task pipeline contains several steps, including entity recognition, relationship extraction, and entity linking. This kind of pipeline method means that errors in any procedure will inevitably propagate to the final prediction. In order to solve the above problem, this paper proposes a Corpus Generation -Retrieve Method (CGRM) with Pre-training Language Model (PLM) and Knowledge Graph (KG). Firstly, based on the mT5 model, we designed two new pre-training tasks: knowledge masked language modeling and question generation based on the paragraph to obtain the knowledge enhanced T5 (kT5) model. Secondly, after preprocessing triples of knowledge graph with a series of heuristic rules, the kT5 model generates natural language QA pairs based on processed triples. Finally, we directly solve the QA by retrieving the synthetic dataset. We test our method on NLPCC-ICCPOL 2016 KBQA dataset, and the results show that our framework improves the performance of KBQA and the out straight-forward method is competitive with the state-of-the-art.

show abstract

“…They combine an attention mechanism with Bi-LSTM to predict the attributes and obtain the final answer. Lai et al [24] adopt word vector similarity and fine-grained word segmentation to map properties. Meanwhile, they use many artificially constructed rules and features to select the correct answers.…”

Section: Related Work a Knowledge Base Question Answeringmentioning

confidence: 99%

Answer Acquisition for Knowledge Base Question Answering Systems Based on Dynamic Memory Network

Fan

et al. 2019

IEEE Access

View full text Add to dashboard Cite

In recent years, with the rapid growth of Artificial Intelligence (AI) and the Internet of Things (IoT), the question answering systems for human-machine interaction based on deep learning have become a research hotspot of the IoT. Different from the structured query method in traditional Knowledge Base Question Answering (KBQA) systems based on templates or rules, representation learning is one of the most promising approaches to solving the problems of data sparsity and semantic gaps. In this paper, an answer acquisition method for KBQA systems based on a dynamic memory network is proposed, in which representation learning is employed to represent the natural language questions that are raised by users and the knowledge base subgraphs of the related entities. These representations are taken as inputs of the dynamic memory network. The correct answers are obtained by utilizing the memory and inferential capabilities. The experimental results demonstrate the effectiveness of the proposed approach.INDEX TERMS Internet of things, human-machine interaction, knowledge base question answering systems, dynamic memory network.

show abstract

Open Domain Question Answering System Based on Knowledge Base

Cited by 21 publications

References 7 publications

Topic enhanced deep structured semantic models for knowledge base question answering

Topic enhanced deep structured semantic models for knowledge base question answering

Calculating Question Similarity is Enough: A New Method for KBQA Tasks

Answer Acquisition for Knowledge Base Question Answering Systems Based on Dynamic Memory Network

Contact Info

Product

Resources

About