ONTOLOGY CONSTRUCTION: BIOINSPIRED IMPROVED SEA LION OPTIMIZATION MODEL FOR SEMANTIC INFORMATION RETRIEVAL
Shital Kakad, Sudhir Dhage · ASEAN Engineering Journal · 2024
The search effectiveness and efficiency completely depend on the ontology design. The objective of information retrieval (IR) is search and retrieves precise and accurate data in response to a user's query. Notably, the existing search engine relies on conventional keywords to search information. It purely compares the user's query with the database and retrieves outcomes without understanding the intended meaning behind the user's query. Therefore, a significant proportion of the outcomes contain unrelated information. Further, designing a new ontology from scratch and evaluating it are challenging tasks. This research work is divided into two decision-making processes (i) data filtering and (ii) data annotation. In this paper, the steps of ontology construction as follows: a) Pre-processing of data b) Implementation of Proposed Jaccard Similarity Evaluation to evaluate the similarity of data c) Data filtering and outlier detection and final step d) Semantic annotation and cluster. The data is filtered by using the evaluated similarity function. Then, the data is grouped separately into wanted data and unwanted data. The unwanted words are called outliers. Based on the semantics, the data annotation is performed and the process of clustering is evaluated for developing the precise cross-domain applications-based ontology. Moreover, the clustering is done based on the similarity evaluation under multiple dictionaries. In the clustering procedure, the optimal centroid selection is considered a challenging crisis. Hence, for solving this issue, this research work widened with the introduction of an Improved Sea Lion (ISnLO), which is the improved version of the Sea Lion Optimization algorithm.