(1+1)-evolutionary gradient strategy to evolve global term weights in information retrieval

Ibrahim, Osman Ali Sadek and Landa-Silva, Dario (2016) (1+1)-evolutionary gradient strategy to evolve global term weights in information retrieval. In: Advances in computational intelligence systems: contributions presented at the 16th UK Workshop on Computational Intelligence, September 7–9, 2016, Lancaster, UK. Advances in intelligent systems and computing (513). Springer, Cham, pp. 387-405. ISBN 9783319465616

Full text not available from this repository.


In many contexts of Information Retrieval (IR), term weights play an important role in retrieving the relevant documents responding to users' queries. The term weight measures the importance or the information content of a keyword existing in the documents in the IR system. The term weight can be divided into two parts, the Global Term Weight (GTW) and the Local Term Weight (LTW). The GTW is a value assigned to each index term to indicate the topic of the documents. It has the discrimination value of the term to discriminate between documents in the same collection. The LTW is a value that measures the contribution of the index term in the document. This paper proposes an approach, based on an evolutionary gradient strategy, for evolving the Global Term Weights (GTWs) of the collection and using Term Frequency-Average Term Occurrence (TF-ATO) as the Local Term Weights (LTWs). This approach reduces the problem size for the term weights evolution which reduces the computational time helping to achieve an improved IR effectiveness compared to other Evolutionary Computation (EC) approaches in the literature. The paper also investigates the limitation that the relevance judgment can have in this approach by conducting two sets of experiments, for partially and fully evolved GTWs. The proposed approach outperformed the Okapi BM25 and TF-ATO with DA weighting schemes methods in terms of Mean Average Precision (MAP), Average Precision (AP) and Normalized Discounted Cumulative Gain (NDCG).

Item Type: Book Section
RIS ID: https://nottingham-repository.worktribe.com/output/818749
Schools/Departments: University of Nottingham, UK > Faculty of Science > School of Computer Science
Identification Number: https://doi.org/10.1007/978-3-319-46562-3_25
Depositing User: Landa-Silva, Dario
Date Deposited: 13 Sep 2016 12:22
Last Modified: 04 May 2020 18:13
URI: http://eprints.nottingham.ac.uk/id/eprint/35587

Actions (Archive Staff Only)

Edit View Edit View