- Marián Šimko and students: Spracovanie textu na FIIT
- Peter Gašpar: Tools for observing emotions
Released prototypes and ongoing experiments
Currently used or experiments in progress:
- ALEF TNG
- ALEF: Adaptive LEarning Framework
- Our Places (NašeObce.sk)
- Hungry student (HladnýŠtudent.sk)
- Personalized Summarizer
- City Lights
- Adaptive proxy
- Zoom-based Navigation
- sme.sk recommender
- Intensity Relationship Analyzer & Presenter
- WebImp: FIIT Website Improver
- Bifrost: Web Search Enhancement
- Trecom: Tree Recommender of News
Marián Šimko, Ivan Srba, Rastislav Krchňavý, Peter Gašpar
Sentigrade is a result of a cooperation of Faculty of informatics and information technologies STU in Bratislava and PR company Seesame. It automatically analyses content published on the social network and calculates sentiment for each comment posted by users. Furthermore, data is also displayed through detailed statistics and graphs.
Alef The Next Generation
Martin Labaj, Matúš Pikuliak, Martin Svrček, Peter Gašpar, Veronika Gondová
AlefTNG (Alef: The Next Generation) is a new – upgraded version of popular e-learning system Alef. New Alef takes further steps to improve learning experience by focusing more on exercises and tests. They are enhanced by several gamification and personalized recommendation techniques.
The main concept remains the same: make learning and verifying knowledge more attractive – both for students and tearchers.
Design and user-interface matches modern concepts and is optimized for use on mobile devices with touch screen. Students are also able to see their current progress and interact with system by giving feedback to content.
Ivan Srba, Rastislav Dobšovič, Marek Grznár, Jozef Harinek, Samuel Molnár, Peter Páleník, Dušan Poizl, Pavol Zbell (experiments in progress)
Askalot is the first educational organization-wide CQA (Community Question Answering) system. It differs from standard open CQA systems (e.g. Yahoo! Answers or Stack Overflow) by its explicit dedication to the educational context. In addition, it differs from existing educational CQA systems by the fact that it involves students across the whole organization while existing solutions involve either an open community or too restricted community of students enrolled for the same class.
Askalot provides students and teachers with: (1) essential functions necessary for question answering, knowledge sharing and collaborative learning; (2) advanced functions for an adaptive support by means of personalized recommendations.
Jakub Ševcech, Roman Burger, Michal Holub, Peter Macko (last change in December 2015)
Annota is a bookmarking service for creating bookmarks and annotations in web pages and PDF files. You can create bookmarks, attach tags to them, highlight text, write comments and share created annotations in groups.
Want to share interesting sites you found? Just create a group, invite your friends or colleagues and have all shared bookmarks in one place. Now you can share your bookmarks and collaboratively annotate them.
The service provides web interface to manage and organize bookmarks and an extension to Firefox web browser for easy interaction with annotations. Use sidebar to bookmark pages, write notes and share them with your groups. Highlight text and write comments to save your thoughts. Add tags to bookmarks for easier search and revision.
Márius Šajgalík (last change in December 2015)
Brumo presents a browser-based distributed multi-agent collaborative user modelling and personalisation platform. Since the main purpose is to enable personalisation of browsed web content to user, Brumo provides an interface for user model access. Via personalisation extensions, user can extend this model even further and based on gathered information about various user characteristics, user can modify content of browsed web pages to work more efficiently and comfortably.
ALEF: Adaptive LEarning Framework
Marián Šimko, Michal Barla, Jozef Tvarožek, Martin Labaj, Ivan Srba, Róbert Móro (last change in June 2016)
ALEF is adaptive web-based educational system built on basic three principles: (1) semantic representation of learning domain, (2) extensible personalization and course adaptation, and (3) student active participation and collaboration during learning. ALEF assembles legacy personalization techniques (adaptive navigation, tailored course presentation) with Web 2.0 key concepts (tagging, sharing, organizing the knowledge) in order to improve student performance and learning experience.
We shift traditional „one-size-fits-all” learning towards collaborative learning personalized according to user actual knowledge estimated by tracking and inferring from student’s actions (based on explicit and implicit feedback) and teacher defined learning goals.
Past ALEF team members: Pavel Michlík, Vladimír Mihál, Maroš Unčík, Roman Burger, Jakub Ševcech, Máté Fejes, Andrea Šteňová
Jozef Tvarožek, started as dissertation project
Peoplia is a collaborative learning system in which a socially intelligent tutoring agent assists students with learning using pseudo-tutor assessments. The system features free-text answering, personalized question generation, and adaptive question selection. Our experiments are directed at assessing: (1) how having a computer “as a friend” can improve motivation and learning; and (2) how the addition of a socially intelligent tutoring friend affects students’ in-class motivation and out-of-class system use.
Ondrej Proksa, started as activity of bachelor student in 2011
The goal is to display all publicly available information about places, villages, districts, regions and organizations in Slovakia. Using graphs and maps it displays statistics of self-government well arranged and thus creating complex tool for comparison and analysis.
Project cooperates directly with the Local Government Development Center. Slovak cities and places are the other partners of this project. All information at NašeObce.sk is publicly available. We gained data from the following publicly accessible government portals.
It works on Ruby On Rail framework. Data are regularly downloaded from publicly available sources.
Ivan Srba, master project (last change in June 2012)
PopCorm, Popular Collaborative Platform, is a collaborative environment developed to support effective collaborative learning. PopCorm employs our method for creating dynamic short-term groups. The proposed method is based on students’ various personal and collaborative characteristics. Students in the created groups are able to communicate and collaborate with the help of four collaborative tools which are suitable for task solving in the domain of collaborative learning: a text editor, a graphical editor, a categorizer, and a semi-structured discussion. The categorizer is a special tool developed for solving different types of tasks the solution of which consists of one or more lists (categories). The semi-structured discussion represents a generic communication tool independent of a particular type of a task being solved.
Róobert Móro, master project (last change in June 2012)
Personalized summarizer generates text summaries by extracting sentences conveying important information from the original documents. For this purpose it uses a method of personalized summarization that unlike the conventional (generic) methods considers other useful information besides the document content, such as a particular user’s characteristics (e.g. her actual knowledge) or her personal annotations (highlights). It utilizes the relevant terms from the domain conceptualization as well.
The summarizer is implemented as a web service and is capable of summarizing texts in many languages independently of the chosen domain. We have integrated it with the adaptive web-based educational system ALEF where it provides summaries of educational texts for students. There, it can be used to help them decide whether the educational text is worth reading or to help them revise before test or exam.
Try Personalized Summarizer in ALEF now!
Peter Dulačka, bachelor project (last change in June 2012)
CityLights is a web game, where players (by their actions) validate sets of music annotations fetched from public datasets. They are presented with couple of sets of annotations and a song. Their task is to guess which of the presented sets contains annotations for song they are hearing. The goal is to go through pregenerated path with minimum of wrong guesses. Players by their actions unknowingly validate individual song-annotation relationships and help us to get rid of unsuitable or too subjective annotations.
Balázs Nagy, bachelor project (last change in December 2011)
PexAce, as a game with a purpose, is a variant of the popular memory game Pexeso where players need two find two same cards on the playing board by subsequently uncovering pairs of cards. In PexAce, we allow players to enter short textual annotations to cards, which help them recall card positions and thus finish the game faster and score more points. We process these annotations via multiple player agreement principles and thus contribute to multimedia metadata authoring by acquiring metadata describing the presented images.
<h3 “>Adaptive ProxyMichal Barla, Tomáš Kramár, Jozef Tomek (last change in June 2011)
Adaptive proxy is an enhanced http proxy server, delivering personalized and social surfing experience on the open and “wild” Web. It incorporates open corpus keyword-based user modeling, various approaches to virtual communities detection and personalizing services, which are, for instance, able to improve your web-based searching or navigation within complex web sites. Adaptive proxy is an open-source platform with architecture based on plugins and services. As such, it is an ideal platform for evaluation of bachelor, master and dissertation theses dealing with Semantic, Adaptive and Social Web. It can host your project too!
Jakub Šimko, part of master project (last change in June 2011)
A simple web search word game – players minimize the number of results returned by a search engine using a special format for queries (example: “jaguar -animal -car -cat -company”) consisting of a given task word (“jaguar”) and “minus words” (“-animal -car -cat -company”), which are guessed by the user.
The game logs are further analyzed and a taxonomy of terms is constructed. The key idea of taxonomy construction is that successful game queries have to contain “minus words” related to the task word.
Explore LittleGoogleGame results here: xlsx
HladnýŠtudent.sk (Hungry Student)
Ľuboš Demovič, Martin Lipták, Jakub Kříž, started as activity of bachelor students
HladnýŠtudent.sk is a website (in Slovak) for all hungry students and everyone who wants to choose something good for lunch. It contains daily menus for all major student canteens in Mlynská Dolina and some basic information about the canteens like the opening hours and the location on the map.
The application runs on Ruby on Rails framework and the canteens’ homepages are automatically parsed every day to keep all the information updated.
We plan to add more canteens from different regions and the users can suggest their favourite ones using a form at the bottom of the page. We have even more plans for the future, such as a mobile version, meal rating or favourite meal notification.
Memotion: Save and Share Memories and Emotions from Photographed Events
Michal Lohnický, master project (last change in June 2011)
Memotion is innovative and unique photo album visualization, which is used for events recall, recognition and for a navigation through the photo album. We follow the fact that in most cases photography is considered to be a medium to save and share memories and emotions from photographed events. Our visualization intended to enhance positive user experience while browsing photo albums.
Try Photo Album Creation now (video show)!
Try Photo Album Browsing now (video show)!
Ján Suchal et al., doctoral project (last change in September 2011)
In cooperation with largest Slovak online news portal (~1M visits/day) we present collaborative recommendation engine for news and blog articles. Our approach includes novel recommendation techniques such as negative implicit feedback from server logs and linearly scalable top-k nearest neighborhood-based recommendation algorithm. We are currently processing and recommending from a stream of ~200MB raw logs/day.
Try sme.sk recommender now! (zakliknúť “Zobrazovať odporúčania priamo v článkoch”)