Clustering Image Search Results by Entity Disambiguation

Abstract

Abstract. Existing key-word based image search engines return images whose title or immediate surrounding text contains the search term as a keyword. When the search term is ambiguous and means different things, the results often come in a mixed bag of different entities. This paper proposes a novel framework that understands the context and thus infers the most likely entity in the given image by disambiguating the terms in the context into the corresponding concepts from external knowledge in a process called conceptualization. The images can subsequently be clustered by the most likely associated entities. This approach outperforms the best competing image clustering techniques by 29.2% in NMI score. In addition, the framework automatically annotates each cluster of images by its key entities which allows users to quickly identify the images they want.

A Novel Model for Imbalanced Data Classification

“…In recent years, binary classification for imbalanced data has received much attention (Castro and Braga 2013; due to its wide applications in various domains such as mortality prediction, and so on (Zhao et al 2014;Bhattacharya, Rajan, and Shrivastava 2017;Liu et al 2018;Nie et al 2019;Liu et al 2019;Lin et al 2019;Xu et al ). A main characteristic of this problem is that, most of data samples belong to one class while the rest belong to the other.…”