Web Technologies include procedures that are used in order to enhance the services that are offered by the World Wide Web. They include both services that can be presented directly to the users of the World Wide Web and services that are transparent to the end user but help in order to promote the usability of the Web. Web Technologies can be separated to many categories. The Lab of Distributed Systems and Telematics focalizes on Dynamic Processing of Web Content (Information Extraction), Data Mining, Web Content Summarization, Categorization and Personalization, and Web Site Construction. The basis of most of the aforementioned Web Technologies is collecting data and extracting keywords. This is a procedure that includes algorithms that are close to lexicographic analysis of the content. Although this analysis could not be thought as a web technology, it still remains an essential part. Moreover, some algorithms for web technologies require the existence of metadata in order to acquire information about the meaning of the content as lot of information include more than text (multimedia). The languages that are used in our research are basically C++, Java, XML, HTML, PHP, ASP.NET, JSP, MSSQL, MySQL and more.
In the chaotic conditions that occur in the web content with the mass creation of web data, Web Technologies are becoming a trend - more than a tool - in order to help the users of the Internet to their navigation. More specifically, dynamic processing of Web Content is a technology that leads to Information Extraction.
Information Extraction is the procedure of extracting the meaning of a text provided that specific type of metadata and simple data are given to the Information Extraction Mechanism.
The basis of Information Extraction is text analysis and more accurately keyword recognition and extraction. The Lab of Distributed Systems and Telematics has put the focus on services that could be used for keyword extraction and thus we are experimenting on creating Information Extraction Systems.
Another crucial part of Web Technologies is further processing of the information that is extracted from web data. This further processing introduces algorithms, procedures and tools that lead to summarization of large amount of data and automatic categorization. These technologies can be used by mechanisms that present data to the users of the web. Additionally, as technology expands, a need for fewer amounts of data arises especially for the users of small screen devices.
Web Content Summarization can be based on simple keyword extraction or on Information Extraction. In the first case, The Lab of Distributed Systems and Telematics, focuses on locating algorithms that could be both quick and effective in order to produce summaries for web content. Some simple but helpful examples of summarization could be web site summarization, article summarization, construction of tools for small screen devices and more.
Web Content Automatic Categorization is the basis of the new trend of the World Wide Web which is Web Personalization. Some Personalization Techniques are based on the actions of the users when using a device, when visiting a device, or when using a program that utilizes web content. The categorization consists of algorithms and procedures that, based on predefined categories, can detect the similarity of a text with a category and rank it according to the most similar category. Additionally, this similarity issue can be used in order to locate the matching between two texts. Our research analysis is focused on locating algorithms that could use a combination of SVM algorithm and positive and negative terms for each category in order to categorize web content.
Our intention is to create mechanisms for summarization and categorization, which can be used as modules to software, in order to enhance the usability and efficiency.
All the aforementioned technologies introduce techniques that are transparent to the user because this is their intention and nature. What is tangible to the World Wide Web users are web sites and services that are offered through them. The Lab of Distributed Systems and Telematics is doing research on web site construction of dynamic content and thus on technologies that include dynamic web languages. The research is put on constructing web sites focalizing on efficiency and re-usability without excluding the human factor and the graphic user interface.
The Lab of Distributed Systems and Telematics has put the focus on Web Site Personalization which seems to be an essential part of all the dynamic content web sites. The Personalization issue that is implemented by The Lab of Distributed Systems and Telematics is based both on the users' actions and on the combination of all the aforementioned technologies. We believe that this combination can lead to better results for enhancing the usability and results that derive for the construction and use of the personalized portals.
Recently, we have expanded our research of Web Technologies on web services, agents and focused crawlers in order to extend our systems with features that are introduced lately to web technologies and can enhance even more any of our modules, especially the services that are provided and promoted through the construction of dynamic web sites and portals.
During the last 2 years we have been working on the development of a system that will include all the subjects of the research that we have been doing onto web technologies. The created system is called peRSSonal and its scope is to present personalized news to the internet users.
The procedures of peRSSonal include:
- Web crawling
- Text extraction from HTML pages
- Text preprocessing
- Keyword extraction
- Text categorization
- Text summarization
- User personalization
A first demo of the system can be found here. The portal presents articles collected from 6 different major news portals in a personalized way. The user can register to the service in order to personalize the content on his/her needs. More specifically, a user is able to select his/her category preference in order to be presented only with articles from specific categories while the system is able to recognize the exact profile of the user and adapt even more the content on his/her needs in order to present even more accurate results.