TY - GEN
T1 - Template-based Question Answering using Recursive Neural Networks
AU - Athreya, Ram G.
AU - Bansal, Srividya K.
AU - Ngomo, Axel Cyrille Ngonga
AU - Usbeck, Ricardo
N1 - Funding Information:
This paper presents a novel approach for the QA over Linked Data task by converting it into a template classification task followed by a slot filling task. Although earlier template-based approaches have attempted similar solutions, this was the first time (to the best of our knowledge) that recursive neural networks were applied to the template classification task. For completeness, a slot filling approach using an ensemble of the best components for named entity, predicate and class recognition tasks were presented. Our evaluation showed that state-of-the-art neural network techniques such as Long Short Term Memory (LSTM), recursive neural networks, and word embeddings be leveraged for the template classification task. We are aware that our approach has a coverage issue in terms of being bound to the training templates and look forward to mitigating this issue through a finer-grained training process. Acknowledgements. We acknowledge the support of the Federal Ministry for Economic Affairs and Energy (BMWi) project SPEAKER (FKZ 01MK20011A).
Publisher Copyright:
© 2021 IEEE.
PY - 2021/1
Y1 - 2021/1
N2 - Most question answering (QA) systems over Linked Data, i.e. Knowledge Graphs, approach the question answering task as a conversion from a natural language question to its corresponding SPARQL query. A common approach is to use query templates to generate SPARQL queries with slots that need to be filled. Using templates instead of running an extensive NLP pipeline or end-to-end model shifts the QA problem into a classification task, where the system needs to match the input question to the appropriate template. This paper presents an approach to automatically learn and classify natural language questions into corresponding templates using recursive neural networks. Our model was trained on 5000 questions and their respective SPARQL queries from the preexisting LC-QuAD dataset grounded in DBpedia, spanning 5042 entities and 615 predicates. The resulting model was evaluated using the FAIR GERBIL QA framework resulting in 0.419 macro f-measure on LC-QuAD and 0.417 macro f-measure on QALD-7.
AB - Most question answering (QA) systems over Linked Data, i.e. Knowledge Graphs, approach the question answering task as a conversion from a natural language question to its corresponding SPARQL query. A common approach is to use query templates to generate SPARQL queries with slots that need to be filled. Using templates instead of running an extensive NLP pipeline or end-to-end model shifts the QA problem into a classification task, where the system needs to match the input question to the appropriate template. This paper presents an approach to automatically learn and classify natural language questions into corresponding templates using recursive neural networks. Our model was trained on 5000 questions and their respective SPARQL queries from the preexisting LC-QuAD dataset grounded in DBpedia, spanning 5042 entities and 615 predicates. The resulting model was evaluated using the FAIR GERBIL QA framework resulting in 0.419 macro f-measure on LC-QuAD and 0.417 macro f-measure on QALD-7.
KW - Question Answering
KW - Recursive Neural Network
UR - http://www.scopus.com/inward/record.url?scp=85102617112&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85102617112&partnerID=8YFLogxK
U2 - 10.1109/ICSC50631.2021.00041
DO - 10.1109/ICSC50631.2021.00041
M3 - Conference contribution
AN - SCOPUS:85102617112
T3 - Proceedings - 2021 IEEE 15th International Conference on Semantic Computing, ICSC 2021
SP - 195
EP - 198
BT - Proceedings - 2021 IEEE 15th International Conference on Semantic Computing, ICSC 2021
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 15th IEEE International Conference on Semantic Computing, ICSC 2021
Y2 - 27 January 2021 through 29 January 2021
ER -