Recommendation Of Blogs In Elearning

Published Date: 02 Nov 2017

1. Abstract:

E-Learning is an interactive teaching and learning e-environment, where teachers and students communicate about their subject. Information about various subjects is provided as e-learning blogs, based on the user’s interest. The challenge in the highly polluted web environment is to identify the best e-learning blog from the existing various blogs. The identification of the best e-learning blog depends on certain criterion such as considering the previous history and personal interest of the user, word of mouth, the rate of the blog in the web and. In the paper proposed, appropriate algorithms are chosen to incorporate the proposed ideology.

2. Introduction:

With the growing technology, e-learning has shown convenience towards learning due to the abundant available of information in the web. But, often the relevant information was not available as a whole; hence, blog came into picture. A blog is expected to be more interactive and informative to create an intuitive site for user’s ease. Such blogs are plenty in existence.

Now, identifying the relevant and the most desired blog is a challenging issue as it depends on many considerations. Basing on the user’s previous history a decision can be made on the user’s personal short-term and long-term interest. The short-term and the long-term interest can be categorized basing on the amount of time they spend using it. The short-term could be the blogs that is visited by the user occasionally depending on the user’s need for a certain period of time. The long-term interest could be considered depending upon the high frequency of usage of a blog by the user. The word of mouth is one such consideration, where, the comments are analyzed to identify the positive and negative posts. The blogs with high positive comments are often considered. The consideration depends choosing the one’s with the highly ranked posts. The above considerations are filtered at each level and the best posts are published. The optimized level of filtrations gives the appropriate and desired result.

3. Related Works:

Blogs are generally popular in internet systems. Since mobile phones are used by many users, it is better to have a blog in mobile phones. Many problems can be discussed in same sites and many users can interact by solving a problem. In PC, to check a blog the system has to be ON, go online and view the particular webpage. But in mobile phones, the updated document is sent to user mobile phones by describing a summary of 100 words. Ranking and comments is provided. Cost is low compared to PC internet. WAP service is used as a bridge between internet and mobile phones.

3. 1. Method for M-CRS WAP (Mobile Content Recommendation System Wireless Application Protocol) in stepwise manner:

1. Classify the documents (use non-negative matrix factorization algorithm)

2. Select high quality documents across themes

3. Analyze user preference

4. Personalized document recommendation is analyzed

5. Send messages to WAP gateway

6. Encode WAP push message into SMS format

7. Deliver message to mobile phones

It is proven by implementing a case study containing 20000 mobile phone users and examined user preferences for a topic and personalized content recommendation.

3. 2. Method for providing mobile users with blog articles that suite their interest using M-CCS (customized content service on mobile devices):

1. Time sensitive criteria that predict the latest popular blog topics available over the web.

2. Conventional recommender system is a hybrid approach proposed to recommend blog articles based on personalized popularity of topic cluster, item based collaborative filtering, attention degree.

3. CO-RSS retrieves blog contents to derive blog topics

4. Double exponential smoothing method to predict popularity degree of the topic basing on the topic rank.

5. Collaborative filtering: combines the above clustered contents to obtain the user desired blogs.

6. WAP push service is the service that sends the filtered articles to mobile devices.

3. 3. Mining-Text From Student-System Interactions to Recommend Blogs and Papers :

The approach integrates a recommended system that retrieves information and recommends blogs, and the WBLE (Web Based Learning Environment) named MASSAYO (a real time scenario) to help students to gain knowledge. The recommendation is based on user’s profile and interaction with the WBLE. It has the ability to find the relevant topics from the recommended blogs. Intelligent Agent keeps a track of the user interaction and provides the recommended blog for the user. Agent oriented approach combines both papers and blogs for learning resources in one recommendation method.

1) Users use the web 2.0 tool to interact with the WBLE.

2) The Intelligent Agent keeps track of the user interaction.

3) Identifies all the users who need help by analyzing the user interaction.

4) The intelligent agent at the end provides the user with the recommended blog.

Technologies used for the recommender system are

1) Lucene library - index the involved resources.

2) Metadata harvester and a blog crawler – collect resource from the web.

3) Hibernate and sesame frameworks – ontology repository and database repository

3. 4. Theory of 6-degrees:

The blog recommender is a flexible and promising mechanism. It combines "trust models, social relation and semantic analysis". It measures the trustworthiness and reliability of the targets, addresses the social intimacy and similarity of blog behavior in social network and also compares textual similarity of blog article. They follow the "6-degrees" theory to develop the mechanism. It is an network based recommendation system.

Bloggers recommendation network can be linked on an average of 6 degree traversal, except for isolated bloggers.

Methodologies:

Modeling of Trust- based blog network

Gathering the information of the blog using the search algorithms.

Constructs the blog network.

Scoring of web blogs

Identifies the social relations and semantic similarities.

Calculates the trustworthiness and reliability known as trust degree (TR).

Identifies the common links, no. of hyperlinks in common, same tags, comments contributed by same author (SIP).

Neural network – based recommendation mechanism

Performs requester evaluation.

Strength of the relation of the bloggers

3. 5. Customer-driven Recommender System:

The recommendation of blog is using is Customer-driven Recommender System (CRS) followed by collaborative filtering. Here searching process is done on the customer network based on similar neighbors. CRS method is efficient and as accurate as traditional recommender system of Global search mechanism.

Customer Network:

Customer network is created in order to effectively collect the information of the customer who is in the network. The customer model consists of Personal information: finds the personal interest of the customer, Neighbor set: gives the information who are the customers with similar area of interest, Content information: contain the information received from neighbors, Target information: contain collection of customer to whom recommendation has to be given. Whenever the is change or update in the purchase transaction there will be change in the target set and neighbored

Network Formation:

The host customer has neighbor whose interest is similar to that of host. The neighbor set sends the recommendation to the host customer and in turn host customer send recommendation to its target customer. When the host customer purchases a product item, it is disseminated to the host’s target customer. Host information set gets updates due to purchase of new item and there will be new neighbor set gets added to it. If host customer does not purchase a product its neighbor set gets updated when its target customer buy a product.

Proposed Architecture:

To incorporate the issues an architecture is been proposed. The system here, improvises the efficiency in recommendation systems in the blogs.

Fig.: System Architecture

User as a client registers to the e-learning blog, where in search is made depending upon the user’s interest, previous history, word of mouth and rating criterion. Each criterion is considered as a level of filtration and the blog is recommended.

When a user searches on a particular topic, the blog relative to it is displayed. While the user browses a blog an event is generated and the url sent to the server is analyzed through text mining using Document Classification Technology which is saved as previous history. When later searches are made on the similar topic, the user is recommended with the blogs that are filtered under the previous history is displayed. After the first level of filtration, the comments are analyzed and the blog posts are classified based on the positive comments using Collaborative Filtering Principle. Basing on the above filtered blog posts, the rate of them is considered using HITS algorithm and the one with highest rate will be given the higher priority for display.