Semantic analysis of web documents for the generation of optimal content

• A mechanism that delivers quality SEO based on LDA and state-of-the-art SE metrics.
• LSHrank performs well against other state-of-the-art approaches.
• Domain of queries/websites influences the results of LDA.
• LDA settings for small or large scale content production influence our mechanism.
• LDA can produce high quality content optimized for higher SE rankings.

The Web has been under major evolution over the last decade and search engines have been trying to incorporate the changes of the web and provide the user with improved – in terms of quality – content. In order to evaluate the quality of a document there has been a plethora of attempts, some of which have considered the use of semantic analysis for extracting conclusions upon documents around the web. In turn, Search Engine Optimization (SEO) has been under development in order to cope with the changes of search engines and the web. SEO׳s aim has been the creation of effective strategies for optimal ranking of websites and webpages in search engines. Current work probes on semantic analysis of web content. We further elaborate on LDArank, a mechanism that employs Latent Dirichlet Allocation (LDA) for the semantic analysis of web content and the generation of optimal content for given queries. We apply the new proposed mechanism, LSHrank, and explore the effect of generating web content against various SEO factors. We demonstrate LSHrank robustness to produce semantically prominent content in comparison to different semantic analysis based SEO approaches.

Figure optionsDownload as PowerPoint slide