3 CONCLUSION OF ANALYSIS & DISCUSSION AS CONSTITUENTS TO SEAR...

4.3 Conclusion of Analysis & Discussion

as constituents to search queries. The process of extracting The period within (1999-2007) was very rich in QA research phrases was performed with the aid of named-entity (NE) than any other time. This was most likely because of the recognition, stop-word lists, and parts-of-speech taggers. challenging research environment provided by QA tracks of Experimental environment and results the TREC annual conferences within that period. However, The QA system was evaluated using two datasets: the other campaigns such as CLEF and NTCIR still represent a AQUAINT corpus, a 3GB collection of news documents used vital vein for QA research. in TREC-2006; and the other dataset was the web. The Also, most researches in the QA field were somehow datasets used 387 TREC questions with non-empty answers. heterogeneous with respect to their system architecture, The documents in the AQUAINT corpus were indexed using approaches, scope, evaluation metrics, etc. But, on the other Lucene engine. Their experiments utilized automatically and hand researches were mainly concerned with one or more of Journal of Computer Science and Network Security, vol. 9, the three basic components of QA systems: question no. 4, pp. 349-358, 2009. classification, document retrieval and answer extraction, [10] E. Voorhees, "Overview of the TREC 2002 Question which combine techniques from natural language processing Answering Track," in Proceedings of the Text Retrieval (NLP), information retrieval (IR), and information extraction Conference (TREC 2002), 2002. (IE), respectively. [11] B. F. Green, A. K. Wolf, C. Chomsky and K. Laughery, "BASEBALL: An automatic question answerer," in