QA SYSTEM COMPONENTS ANSWERS TO QUESTIONS RATHER THAN FULL D...

2. QA System Components

answers to questions rather than full documents or best- As shown in (Figure 1), a typical QA system consists of three matching passages, as most information retrieval systems distinct modules, each of which has a core component beside currently do. other supplementary components: “Query Processing However, the main type of questions submitted by users in Module” whose heart is the question classification, the natural language are the factoid questions, such as “When did “Document Processing Module” whose heart is the the Egyptian revolution take place?” But, the recent research information retrieval, and the “Answer Processing Module” trend is shifting toward more complex types of questions such as definitional questions (e.g. “Who is the President of whose heart is the answer extraction. Question processing is the module which identifies the focus Egypt?” or “What is SCAF?”), list questions (e.g. “List the of the question, classifies the question type, derives the countries that won the Cup of African Nations”), and why-type questions (e.g. “Why was Sadat assassinated?”). expected answer type, and reformulates the question into The Text Retrieval Conference (TREC), a conference series semantically equivalent multiple questions. co-sponsored by NIST, initiated the Question-Answering Reformulation of a question into similar meaning questions is Track in 1999 which tested systems’ ability to retrieve short also known as query expansion and it boosts up the recall of text snippets in response to factoid questions (for example, the information retrieval system. Information retrieval (IR) system recall is very important for question answering, “How many calories are in a Big Mac?”) [3]. Following the success of TREC, in 2002 the workshops of both the Cross because if no correct answers are present in a document, no further processing could be carried out to find an answer [6]. Language Evaluation Forum (CLEF) and NII Test Collection for IR Systems (NTCIR) started multilingual and cross-Precision and ranking of candidate passages can also affect