Design and Analysis of Curation Algorithm on Blog Posts Using Hybrid Computing

Abstract

Nowadays, blogging and micro-blogging have a high impact on the world wide web. Blogs have captured a significant portion compared to other web tools and services. Individuals write blogs, which give wonderful information on any topic to people all over the world. People now share their thoughts, experiences, and viewpoints with others, this is the reason for the growing popularity of blogging services. However, for the users, retrieving essential information from multiple blogs and the blogosphere is a difficult task.The number of blog users and micro-blogs users has exponentially increased in the last few years. For users, to get relevant blog posts on minimum clicks and minimum time, there is a requirement of two things: Content Curation of blog posts and Recommendation on blog posts. Curation is an important step in locating relevant content for the topics being searched. Many approaches for creating summary content have been presented, however, they have always concentrated on providing accurate content that lacks the main substance of the input texts.The content Curation algorithm will find the blog posts on behalf of personalized search results. Secondly, the blog post recommendation system will help to find the most appropriate match for the user to interact with. newlineIn the thesis, an adaptive and intelligent blog post model is discussed which shows the relevant blog posts based on the user s interest. The thesis work demonstrates a technique that incorporates both curated and suggested results from the user. To create information-rich abstractions, the study focuses on a hybrid model that integrates self-attention with a bi-directional long short-term memory auto-encoder, dubbed the Bi-LSTM-AE Model. After pre-processing the dataset, the primary word-level and sentence-level characteristics are retrieved. The extraction summary is then prepared based on the similarities between the contents and sent to the auto-encoder for final abstraction.