Incorporating abstract generation into an online retrieval interface for a library newspaper cutting system

The newspaper is a very widely used medium for transmitting news and information in today's society. Some of the published information is very important and therefore required to be kept for future reference. The cutting and subsequent retrieval of newspaper articles is a tedious process. The retrieval process is also very laborious. Electronic newspaper cutting is required to replace the manual process. A library newspaper cutting system was developed at Nanyang Technological University, Singapore. The system enables librarians to cut, index and store news articles interactively. Readers can retrieve news articles via World Wide Web (WWW) browsers. Different abstract generation techniques, including location method, indicative‐phrases, keyword frequency and title‐keyword method, are incorporated into the retrieval interface to filter out irrelevant news articles. In this paper, different abstract generation techniques and their effectiveness as a means to generate indicative abstracts for news articles are discussed.