网络学术文献内图片的自动识别与检索

Automatic Identification and Retrieval of Images from Academic Literatures on the Internet

学术性文献内的图片具有确定信息需求、辅助滤检的功能,网络学术文献内的图片在上下文语境、大小、命名、格式、超链接等方面有着独特的特征,可以与网络广告、图标等干扰图片区分开来,便于实现自动识别和辅助检索。基于这样的理论,以Google的自动抓取为基础采用VC++进行系统开发,实现了对含图片的文献的检索并在结果页面上以缩略图的形式展示图片,并依此对检索重新排序。

Images in academic literatures have the role of identifying the information demand and assisting retrieval.Images in network academic literatures can be distinguished from online advertisements and icons for their unique features in the context of text,size,name,format,hyperlink and so on so that automatic identification and assistant retrieval can be achieved.Based on this theory,we use Visual C++,which is based on Google’s automatic crawl,to develop the system,realize the retrieval of literatures containi...