Vassilis Poulopoulos

Platform for distributed 3D gaming

Video games are typically executed on Windows platforms with DirectX API and require high performance CPUs and graphics hardware. For pervasive gaming in various environments like at home, hotels, or internet cafes, it is beneficial to run games also on mobile devices and modest performance CE devices avoiding the necessity of placing a noisy workstation in the living room or costly computers/consoles in each room of a hotel. This paper presents a new cross-platform approach for distributed 3D gaming in wired/wireless local networks. We introduce the novel system architecture and protocols used to transfer the game graphics data across the network to end devices. Simultaneous execution of video games on a central server and a novel streaming approach of the 3D graphics output to multiple end devices enable the access of games on low cost set top boxes and handheld devices that natively lack the power of executing a game with high-quality graphical output.

Personalized news categorization through scalable text classification

Existing news portals on the WWW aim to provide users with numerous articles that are categorized into specific topics. Such a categorization procedure improves presentation of the information to the end-user. We further improve usability of these systems by presenting the architecture of a personalized news classification system that exploits user’s awareness of a topic in order to classify the articles in a ‘per-user’ manner. The system’s classification procedure bases upon a new text analysis and classification technique that represents documents using the vector space representation of their sentences. Traditional ‘term-to-documents’ matrix is replaced by a ‘term-to-sentences’ matrix that permits capturing more topic concepts of every document.

An efficient mechanism for stemming and tagging: the case of Greek language

In an era that, searching the WWW for information becomes a tedious task, it is obvious that mainly search engines and other data mining mechanisms need to be enhanced with characteristics such as NLP in order to better analyze and recognize user queries and fetch data. We present an efficient mechanism for stemming and tagging for the Greek language. Our system is constructed in such a way that can be easily adapted to any existing system and support it with recognition and analysis of Greek words. We examine the accuracy of the system and its ability to support peRSSonal a medium constructed for offering meta-portal news services to internet users. We present experimental evaluation of the system compared to already existing stemmers and taggers of the Greek language and we prove the higher efficiency and quality of results of our system.

Efficient extraction of news articles based on RSS crawling

The expansion of the World Wide Web has led to a state where a vast amount of Internet users face and have to overcome the major problem of discovering desired information. It is inevitable that hundreds of web pages and weblogs are generated daily or changing on a daily basis. The main problem that arises from the continuous generation and alteration of web pages is the discovery of useful information, a task that becomes difficult even for the experienced internet users. Many mechanisms have been constructed and presented in order to overcome the puzzle of information discovery on the Internet and they are mostly based on crawlers which are browsing the WWW, downloading pages and collect the information that might be of user interest. In this manuscript we describe a mechanism that fetches web pages that include news articles from major news portals and blogs. This mechanism is constructed in order to support tools that are used to acquire news articles from all over the world, process them and present them back to the end users in a personalized manner.

CUTER: An Efficient Useful Text Extraction Mechanism

In this paper we present CUTER, a system that processes HTML pages in order to extract the useful text from them. The mechanism is focalized on HTML pages that include news articles from major portals and blogs. As useful text we define the body of the article that contains the news report. In order to extract the body of the article we deconstruct the HTML page to its DOM model and we apply a set of algorithms in order to clean and correct the HTML code, locate and characterize each node of the DOM model and finally store the text from the nodes that are characterized as useful text nodes. CUTER is a subsystem of peRSSonal, a web tool that is used to obtain news articles from all over the world, process them and present them back to the end users in a personalized manner. The role of CUTER is to feed peRSSonal with the body of the. In this paper we present the basic algorithms and experimental results on the efficiency of the CUTER text extractor.

Large scale opinion mining for social, news and blog data

A business intelligence platform with social basis.Real-time data filtering in the source.Summarization of historical content.Statistics computation over sliding windows. Companies that collect and analyze data from social media, news and other data streams are faced with several challenges that concern storage and processing of huge amounts of data. When they want to serve the processed information to their customers and moreover, when they want to cover different information needs for each customer, they need solutions that process data in near real time in order to gain insights on the data in motion. The volume and volatility of opinionated data that is published in social media, in combination with the variety of data sources has created a demanding ecosystem for stream processing. Although, there are several solutions that can handle information of static nature and small volume quite efficiently, they usually do not scale up properly because of their high complexity. Moreover, such solutions have been designed to run once or to run in a fixed dataset and they are not sufficient for processing huge volumes of streamed data. To address this problem, a platform for real-time opinion mining is proposed. Based on prior research and real application services that have been developed, a new platform called PaloPro is presented to cover the needs for brand monitoring.

Enhancing meta-portals using dynamic user context personalization techniques

The Internet is flooded with information and the last decade its size has grown so many times that information search and presentation have become tedious tasks even for experienced users. Minor changes to existing resources can alter the situation and lead to major changes to the end user experience. In this manuscript we present the dynamic web personalization and document grouping infrastructure for meta-portals and the evaluation of our mechanism on a meta-portal. A meta-portal is an informational node where articles from different sources are collected and presented in a categorized and personalized manner. The web personalization mechanism is based on dynamic creation and update of user profiles according to the users preferences when browsing. In parallel a users profile is affected by user grouping details, which are constructed by users with similar profiles. Assuming that required information, such as article tagging, keywords to categories matching and articles to categories relation is already part of the meta-portal we present a novel mechanism that can build and maintain a user profile which is formed without disturbing the user. Furthermore, we describe the real-time user-centred document grouping mechanism that is implemented to support the web personalization system and present the experimental evaluation of the whole system.

Utilizing RSS Feeds for Crawling the Web

We present “advaRSS” crawling mechanism which is created in order to support peRSSonal, a mechanism used to create personalized RSS feeds. In contrast to the common crawling mechanisms our system is focalized on fetching the latest news from the major and minor portals worldwide by utilizing their communication channels. The challenge between “advaRSS” and a usual crawler is the fact that the news is produced in a random order any time of the day and thus the freshness of the offline collection can be measured even in minutes. This means that the system has to be updated with news every single time they occur. In order to achieve this we utilize the communication channels that exist on the modern architecture of the WWW and more specifically in almost every modern news portal. As the RSS feeds are used by every major and minor portal it is possible to keep our crawler up to date and retain a high freshness of the “offline content” that is maintained in our system’s database by applying algorithms in order to observe the temporal behaviour of each RSS feed.

PaloPro: a platform for knowledge extraction from big social data and the news

PaloPro is a platform that aggregates textual content from social media and news sites in different languages, analyses them using a series of text mining algorithms and provides advanced analytics to journalists and social media marketers. The platform capitalises on the abundance of social media sources and the information they provide for persons, products and events. In order to handle huge amounts of multilingual data that are collected continuously, we have adopted language independent techniques at all levels and from an engineering point of view, we have designed a system that takes advantage of parallel distributed computing technologies and cloud infrastructure. Different systems handle data aggregation, data processing and knowledge extraction and others deal with the integration and visualisation of knowledge. In this paper, we focus on two important text mining tasks, named entity recognition from texts and sentiment analysis to extract the sentiment associated with the corresponding identified entities.

Adaptation of RSS feeds based on the user profile and on the end device

In the last decade, the advances in technology along with the ease of access to information have dramatically changed the World Wide Web status during the last few years. The Internet acts as a means of finding useful information and more specifically news articles. Additionally, more and more people want to utilize their mobile devices towards the scope of reading news articles. The aforementioned situation generates a significant, yet almost untouched problem: easily locating interesting news articles on a daily basis within the space that is available on the small screen device. In our work, we propose a framework that, by utilizing RSS feeds, is able to personalize on the needs of the users and on the capabilities of their device, in order to present to them only a fraction of the news articles and merely the useful information that derives from them. Deploying a generalized, multi-functional mechanism that produces efficient results for the situation described, seems to be a panacea for most of the text-based, information retrieval needs. Within this framework we created PeRSSonal, a mechanism that is able to create personalized, pre-categorized, dynamically generated RSS feeds focalized on the end users small screen device. The system is based on algorithms that incorporate the user into the categorization and summarization procedures, while the articles are presented back to him/her according to her interests and the client device capacity.