A Chinese Automatic Text Summarization system for mobile devices

A Chinese Automatic Text Summarization system for mobile devices Lei Yu, Mengge Liu, Fuji Ren, Shingo Kuroiwa Faculty of Engineering, The University of Tokushima 2-1 Minamijosanjima, Tokushima 770-8506, Japan School of Information Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China {yulei, liumengge, ren, kuroiwa}@is.tokushima-u.ac.jp Abstract. A large amount of on-line information and lengthiness information can’t fit for the mobile devices. In order to save this problem, we propose a method which collects original news text from on-line information and extracts summary sentences from them automatically. On this basis, we adopt WML(Wireless Markup Language) to build a news website for mobile devices browsing through the news summary. The system is mainly made up by Automatic News Collection and Auto Text Summarization. Our experimental results proved the effectiveness of the means. Keyword: World Wide Web; Automatic Text Summarization; Automatic News Collection 1 Introduction As the news websites grow rapidly, the on-line news becomes large and repeatedly, even some news websites quote the contents of other news websites[1]. It is more and more difficult for the reader to skim over websites and get a quick idea of their content. Therefore, based on reporting intensity and updating speed, we select several Chinese news Home sites as the resource of news collection. In the last few decades, researchers have brought up a number of automatic summarization methods[2][3][4][5][6]. Basically these methods fall into two categories. One is called human-imitation approach or generative approach, which constructs the summary based on the structure and semantics of the source text. The other group of approaches tries to extract important sentences based on information acquired from the surface clues of the text. This approach is termed the extraction approach. Therefore, at the current stage, it is better to use statistical information in automatic extraction. The method proposed in this paper also extracts important sentences based on textual statistical information. But our method not only uses statistical information but also employs text structural features to improve the performance of important sentences extraction. 2 Basic idea of Automatic News Collection The basic approach of Automatic News Collection is to analyze target websites and find out the rule of the articles text in HTML page. We have built the experiment news collection system named ACnews to automatically collect news on the proposed method. In this section, we describe the main process for extracting news content from a website. (1) Download a news HTML page from destination websites. (2) Analyze the HTML page to find out the rule. (3) Collect news HTML pages from destination websites. (4) Extract information about news author and news date. (5) Extract news text from HTML pages. (6) Delete ad and ad link from original news text. 426 (7) Save the information obtained by fourth step and fifth step in database. At present, ACnews is only for Chinese articles collection. Of course our method can be applied to other language articles easily, just by additional support of the multinational language code. 3 Basic idea of Automatic Text …

Content truncated to 500 words