The Research And Implementation Of Method For Multi-modal Information Retrieval

With the development of internet, the human society has stepped into information area, at the same time, not only the single user, but also the company and other unions, has a rapid growth for the information demand, so from some extend, it can’t meet the demand of the all humans. In the past, information is only for single retrieval. that is, for text input, the result if just for relative text, and for image, the result is just for relative image, but now, it is the time of text and image, the single modal does not satisfy the demand of the user, that is the cause of emerging of the multiple modal retrieval, you can search the text through image, or get the image through text, thus fully riches the content of the whole industry of information.This article mainly focus on the specified methods about multi-modal information retrieval, including pre-process of image and text, how to build the heterogeneous network, and how to model the text and image with LDA and SIFT algorithm, even how to use Random Walk to query information in a network. At last, this article points out the existing question and comes up with effective method to deal with it.