MIWE: Multimodal Indexing of Web Entities Incorporating Semantic Artificial Intelligence

Web indexing is of critical importance to Web 3.0 owing to the reason that it is not just expanding but growing exponentially. This paper proposes a multimodal indexing mechanism that indexes web images, videos, pages, and documents. The strategy encompasses paradigms like TF-IDF, ontology generation, and label extraction from different datasets. The datasets are subjected to the LSTM for classification for the web image and web video datasets and an LR classifier is used to classify the web document and web page dataset. Subsequently, the formalization of knowledge graphs and enhancement using Google’s knowledge graph API, context frees, microformat chain derivations, the encompassment of Wikidata and NELL knowledge stores, and web crawler agents increases the quantity of auxiliary knowledge, and the convergence to optimality. Semantic relevance is calculated using the NPMI measure. The MIWE shows improved performance when examined against other similar models.