Abstract
In this paper, we describe the web information management system with evaluation results. This will include the usage of the classification system. In the following pages there is relevant research and description as to experiment design, evaluation criteria, and results. Finally I should discuss important findings and implications of the research as conclusion.
1. INTRODUCTION
arrive at a classification that is going to solve requirements of all applications. A
Web classification strategy is easily trainable with different types of class definitions.
One of the most useful applications of Web classification approaches is called Hierarchical Classification system. In this there are taxonomic structures used to organize large amounts of documents. The most typical examples of hierarchical classifications are file systems, market places catalogs and the directories of web portals. According to the above picture all web pages are related according to its content and have been sub divided into several classes. At this point this is referred this groups can be matched and grouped and can be called parent layer.
1.1 Review
Buyukkokten et al presents a browsing system, power browser, which allows transforming of web page into a hierarchy of text blocks, ‘called semantic textual units’. This excluded images and formatting information. In principle these technique could be reused to implement an alternative transformer such techniques do not keep the original look and feel. Hwang et al make minor improvements to the previous techniques. This involves splitting a page into an index page and an individual page for each of them. To detect repeated layout patterns, it uses a grouping function which detects items that structurally share a common prefix that is, items starting with the same sequence of HTML tags.
2. Experiment Design Description Evaluation criteria
The experiment given by our professors, in general requires us to put about 2000 pages of contents into a number of groups. Categorizing is often posed as a supervised learning setback in which a set of labeled data is used to instruct a classifier which can be applied to label future examples. This online system would initially create a database with folders which could be further created by means of adding a new column under each of the main class. In the classification process students have to be quite familiar with the application and utilization of this system. In this case classification can be divided into binary classification and multi class classification, where binary classification categorizes instances into exactly one of two classes positive or negative while multi class classification deals with more than two classes. The task of reading 2000 pages one by one is a magnitude task of great enormity. In this case the use of the recommendation classification method is essential where we are allowed to enlist a rule where we use a list of key words based on the content. The web classification methods can be used to group different with the same content. In the experiment the use of recommendation classification enlisted key words that helped in classification.
2.3 Results
In the results generated, recommendation classification brings forth accuracy in the number of pages on each alignment. In using of the binary and multi class classification in the experiment brings out a higher portion on the positives and negative effects related to web classification. Why this maybe so is normally caused by the overlapping of applications within the formatted settings of web classification tools. In most cases, the overlaps of certain range of key words may cause significant distractions which normally happened after a certain number of rules have been established. This gives rise to a hardship and essential disruption in relation to the classification. In the experiment it was realized that the pages grouped in one set became a challenge and it was seen that to come to such results required both precision and technological knowledge.
3. Personal Feelings
The use of the web classification methods has been revealing in many ways. Though there are shortfalls that we may not be able to tackle with classification. I still believe that the use of web classification makes it easier for categorizing of content on the web. In the experiment what stood out was the simplicity yet complicated systems used to make information more accessible. I believe we should continue researching for the appropriate technology.
