Improving contextual advertising matching by using Wikipedia thesaurus knowledge

Short text categorization exploiting contextual enrichment and external knowledge

“…In other proposals, Wikipedia is exploited to compute semantic relatedness between words or texts, like in [11], and more recently to identify the word sense with a disambiguation process, as described in [4]. Another recent use of Wikipedia knowledge is to enrich the semantic expression of a target commercial advertisement, as presented by Xu et al in their work on contextual advertising [10].…”

“…Several works in the literature exploit external sources to enrich the original set of words with other additional words, in order to add semantic value to the text and improve the categorization process; they will be described in the next section of this paper. Many of these approaches [1,3,4,5,6,8,10,11] focus on semantic and syntactic analysis in order to better detect the meaning of words and phrases by solving problems such as synonymy or polisemy, but giving not so much importance to the temporal context, i.e. when a sentence is made.…”

Constructing plausible innocuous pseudo queries to protect user query intention

“…Moreover, it contains a hierarchical categorization system, where each concept or category belongs to at least a parent category. All these features enable it to be exploited as a knowledge ontology for understanding query intention and capturing query characteristics [26] (see [27,28] for more detail on the structure of Wikipedia).…”

An efficient Wikipedia semantic matching approach to text document classification

“…In addition, there are some methods proposed to improve the TF-IDF model [14]. However, on account of only considering surface text information and ignoring semantic information, the keyword matching techniques would lead to such problems as semantic confusion caused by polysemy and content mismatch caused by synonymy, resulting in a negative impact on the effectiveness of this kind of techniques [12,33,36].…”

“…In general, traditional document classification algorithms were developed based on keyword matching [18,13], whose basic idea is to represent a document as a vector of weighted occurrence frequencies of individual keywords, and then analyze the relevance of keyword vectors to measure the text similarity of documents. However, keyword matching techniques only take into consideration the surface text information, and do not consider the semantic information contained in documents, resulting in problems such as semantic confusion caused by polysemy, and content mismatch caused by synonym, thus reducing the effectiveness of the techniques [12,33,5]. To solve this problem, a new technique called Wikipedia matching was proposed [10,11,3,1], whose basic idea is to use the semantic concepts from Wikipedia as an intermediate reference space, upon which a document is mapped from a keyword vector to a concept vector, so as to capture the semantic information contained in the document.…”