Certain investigations on web image re ranking and retrieval using machine learning techniques

Abstract

The internet has completely revolutionized the world in business transactions, source of information, communication and socialization among others. It has contributed immensely to the nation economic growth and has greatly influenced the exponential growth of computer network users. However, the internet usability increase is inherently bogged with information systems. Web search engines mostly use keywords as queries and rely on surrounding text to search images. They suffer from the ambiguity of query keywords, because it is hard for the users to accurately describe the visual content of target images using only keywords. Most of the existing re-ranking methods utilize the visual information in an unsupervised and passive manner to overcome the semantic gap which is the gap between the low level features and high-level semantics. Although multiple visual modalities have been used to further mine useful visual information they can only achieve limited performance improvements. This is because these re-ranking approaches neglect the intent gap which is the gap between the representation of usersand#8223; query/demand and the real intent of the users. Usersand#8223; real search intent is hard to measure and capture without the user participation and feedback. Some researchers, therefore, attempt to integrate usersand#8223; interaction with the search process. However, it is not easy to obtain sufficient and explicit user feedback since users are often reluctant to provide enough feedback to search engines. newlineTo achieve higher efficiency, the visual feature vectors need to be short and their matching needs to be fast. However, these approaches are only applicable to low dimensional image sets of relatively small sizes, but not suitable for online web-scale image re-ranking with higher dimensional space. newline