Efficient Multi-Document Summary Generation Using Neural Network

Ms Sonali Igave • Prof C. M. Gaikwad

Abstract

From last few years online information is growing tremendously on World Wide Web or on user's desktops and thus online information gains much more attention in the field of automatic text summarization. Text mining has become a significant research field as it produces valuable data from unstructured and large amount of texts. Summarization systems provide the possibility of searching the important keywords of the texts and so the consumer will expend less time on reading the whole document. Main objective of summarization system is to generate a new form which expresses the key meaning of the contained text. This paper study on various existing techniques with needs of novel Multi-Document summarization schemes. This paper is motivated by arising need to provide high quality summary in very short period of time. In proposed system, user can quickly and easily access correctly-developed summaries which expresses the key meaning of the contained text. The primary focus of this paper lies with thef_β-optimal merge function, a function recently presented here, that uses the weighted harmonic mean to discover a harmony in the middle of precision and recall. Proposed system utilizes Bisect K-means clustering to improve the time and Neural Networks to improve the accuracy of summary generated by NEWSUM algorithm.

Metrics

  • 68 views
  • 16 downloads

Journal

International Journal of Advanced Engineering, Management and Science

The International Journal of Advanced Engineering, Management and Science (IJAEMS) is an internat... see more