Ambiguity and Disambiguation in Information Retrieval Systems (Systematic Review)

Ambiguity arises when more than one meaning and concept can be understood from a word, phrase or sentence. Since it seems necessary to understand this by the information retrieval system in order to increase the accuracy of the information retrieval system and increase the retrieval of related resources, the present research aims to identify the ambiguous and disambiguating factors through a systematic review of studies. It has been done in Iran.

The research method is applied in terms of purpose, in terms of approach, qualitative and in terms of information gathering method, systematic review using PRISMA standard. The statistical population includes journal articles, conferences, and dissertations indexed in Iranian databases, including: Magiran, Noormags, Shiraz Regional Center for Science and Technology, Civilica, and Academic Jihad Center (SID) and Scientific Information of Iran (treasure). 175 scientific sources and 138 scientific sources were excluded based on the output criteria and 37 scientific sources were selected based on the input criteria. Input criteria: focusing on the studies conducted on the subject of ambiguity and disambiguation, in the fields of linguistics, information technology, artificial intelligence, computer and information science, and epistemology, scientific sources published in Persian language, without publication time limits, review articles, Scientific research, master's and doctorate theses and national and international conference papers held inside Iran. Output criteria: studies carried out except for ambiguity and disambiguation, non-Persian language scientific sources, books, reports, editorials, abstract writings and short articles (less than 5 pages)

The results show that the main topic of interest in this research was "ambiguity of the meaning of words" (53%) and "disambiguation of words" (44%) and the least topics of ambiguity in machine translation (21%) and disambiguation through ontology (5 percent). The most research methods with the "neural network" approach (22 percent) and the least research methods used were content analysis, methodology and machine learning (5 percent), the largest number of sources related to the articles of conferences and national conferences and was international (48 percent) and the least were master theses and doctoral dissertations (5 percent). Computer and linguistics fields (27 percent) have had the most scientific productions in this regard, and information science and epistemology fields (9 percent) have had the lowest productions.

The present research is considered innovative in the sense that it deals with the issue of ambiguity and disambiguation in information retrieval with a systematic review approach. Its findings indicate that most Iranian researches were focused on disambiguation, which were mostly presented in national and international conferences and meetings. The topic of ambiguity and disambiguation is more important in the fields of computer science and linguistics, and the fields of information science and epistemology have not dealt with it. The semantic ambiguity of words and, by nature, the disambiguation of words has been the concern of format in Iranian researches. Ontology as a new approach to disambiguation has received less attention.