Authors : Ghaidaa A. Al-Sultany and Suha Kamal
Abstract: Latent Dirichlet analysis has emerged as an efficient research field for dealing with the noisiness, high dimensional issues of the text semantically. Enhancing the LDA performance using lexical semantics has proposed in this study. It has focused on enriching the annotated text descriptions of given images with the meaning of the lexical terms with their syntaxes. The semantic lexicon WordNet was applied to produce additional synonym’s terms for the images. The aggregated text features with their synonym have been fetched to the LDA to extract enriched topics. The proposed system was evaluated in terms of the precision, recall and F-measure evaluation measures. The finds out have been discussed and compared against the results of the method without the semantic lexicon support. The proposed research was tested on the benchmark CLEF dataset, the output topics have shown a significant enhancement with respect to the retrieved images.
Ghaidaa A. Al-Sultany and Suha Kamal, 2017. Annotated Images Retrieval using Semantic Analysis and Topic Modeling. Journal of Engineering and Applied Sciences, 12: 8490-8497.