Rating is available when the video has been rented.
This feature is not available right now. Please try again later.
Published on Oct 24, 2016
Kun Xu: Knowledge Based Question Answering
Abstract: As very large structured knowledge bases have become available, answering natural language questions over structured knowledge facts has attracted increasing research efforts. We tackle this task in a pipeline paradigm, that is, recognizing users’ query intention and mapping the involved semantic items against a given knowledge base (KB). we propose an efficient pipeline framework to model a user’s query intention as a phrase level dependency DAG which is then instantiated regarding a specific KB to construct the final structured query. Our model benefits from the efficiency of structured prediction models and the separation of KB-independent and KB-related modelings. The most challenging problem in the structure instantiation is to ground the relational phrases to KB predicates which essentially can be treated as a relation classification (RE) task. To learn a robust and generalized representation of the relation, we propose a multi-channel convolutional neural network which works on the shortest dependency path. Furthermore, we introduce a negative sampling strategy to learn the assignment of subjects and objects of a relation.
Though knowledge based question answering systems can precisely answer some factoid questions, due to the incompleteness and imperfection of the KB, they will still fail at answering many questions. Fortunately, we find external textual sources such as Wikipedia can offer additional evidence to improve both the question coverage and overall performance of a KB-QA system. Specifically, we propose two methods to incorporate the free text into the KB-QA system. The first one is in a pipeline fashion where we additionally perform the text based inference after the traditional KB based inference. The second one is to employ a joint inference model to simultaneously understand the query intention both from the KB and text. Experiments show that these two methods achieve the state-of-the-art performances on two benchmark data sets.