Implementing clarification dialogues in open domain question answering

Abstract

We examine the implementation of clarification dialogues, a mechanism for ensuring that question answering systems take into account user goals by allowing them to ask series of related questions either by refining or expanding on previous questions with follow-up questions, in the context of open domain Question Answering systems. We develop an algorithm for clarification dialogue recognition through the analysis of collected data on clarification dialogues and examine the importance of clarification dialogue recognition for question answering. The algorithm is evaluated and shown to successfully recognize the start and continuation of clarification dialogues in 94% of cases. We then show the usefulness of the algorithm by demonstrating how the recognition of clarification dialogues can simplify the task of answer retrieval.

Semantic similarity based on compact concept ontology

“…However, we found that a straightforward extension, e.g. [1], did not work well in our project on context-based video retrieval, where the videos are typically accompanied by text descriptions such as google video. To achieve better performance, we propose a new method of computing semantic similarity between two articles based on WordNet [2], which is a concept ontology having a tree structure as shown in Figure 1(a) (See Section 2).…”

“…A straightforward method is given in [1] to combine the semantic similarity between two words to compute the semantic similarity between two articles. Given two articles P and P , their semantic similarity is computed as follows in [1] …”

“…is the semantic similarity between two words computed by the method [4] that is based on WordNet, wj and w i are words in articles P and P , respectively. The algorithm in [1] treats each word pair independently, and thus cannot capture the semantic of the article as a whole. For example, a word may have multiple senses and the context will help to identify the real sense in an article.…”

“…Compared with the method in [1], we compute similarity as a whole. In addition, our method is more efficient.…”

“…Suppose the depth of ontology tree and number of tree nodes are constant. The time complexity of the algorithm in [1] is O(m * n), where n and m are the number of words in two articles; The time complexity of our algorithm is O(m + n), either for calculating the similarity or building the weighted tree.…”