Seminar on Informatics and Methodology
Kristina Ban, Faculty of Social Sciences, Faculty of Information Studies, Novo mesto, Slovenia
A short review of network aligners
Wednesday, 6.5.2015, at 14:00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
Among the most challenging problems in network analysis is the comparison of two networks, i.e., the quantification of their topological difference. The challenge comes from this being an NP-hard problem, due to the underlying subgraph isomorphism. Thus, we rely on heuristics through different comparative methods, such as network integration, network querying and the most common one, network alignment. In this talk we will present a short survey of network alignment.
dr. Tomaž Urbič, Faculty of Chemistry and Chemical Technology, University of Ljubljana, Ljubljana, Slovenia
Computer demanding methods in chemistry and chemical technology
Friday, 27. 3. 2015, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
In theoretical chemistry and chemical engineering the use of powerful computers and advances algorithms is key thing to obtain results for complex systems. Computing methods like ab initio method in quantum chemistry, molecular dynamics and Monte Carlo simulation in protein folding and docking, and Lattice Boltzmann method for flow of matter are very time consuming and to get results one has to run code in parallel in multiple computers or GPU in parallel mode. Basics of these methods will be presented and demonstrated on simple problems.
Jože Bučar, Faculty of information studies, Novo mesto, Slovenia
Sentiment based classification of the web texts
Wednesday, 17. 12. 2014, at 10.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
An enormous quantity of data is generated on the web daily. From this chaotic cluster of data we strive to obtain valuable information, which may significantly impact strategic decisions of both business and individuals in the future. The increasing interest in web content has attracted the collaboration among scientists from various fields such as computer science, data mining, machine learning, computational linguistics, graph theory, neural networks, sociology, and psychology.
Within our research we retrieved all textual documents such as news articles from the digital archive of five popular Slovenian websites enriched with political, business, economic and finance content between September 1st 2007 and December 31st 2013. By labelling a sample of approximately 10.000 documents we obtained a labelled corpus, which was used as a training set to train, test and evaluate classification techniques. In this talk, we will present latest results of document classification according to various mining techniques and performance.
dr. Primož Južnič, Faculty of Arts, University of Ljubljana, Ljubljana, Slovenia
Uporaba različnih informacijskih virov pri ocenjevanju znanstvene odličnosti
Thursday, 27. 11. 2014, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
Zagotovo ni družbene dejavnosti, ki bila tako odvisna od informacijskih virov, kot je to znanstveno raziskovanje. Informacijski viri v znanosti imajo dvojen značaj:
1. So osnova znanstvenega informiranja in komuniciranja.
2. Igrajo ključno vlogo v sistemih ocenjevanja znanstvene odličnosti in odmevnosti.
Evalvacija raziskovanja je postala pomembna naloga upravljalcev raziskovalne dejavnosti in nosilcev raziskovalne politike. V Sloveniji ta poteka s pomočjo sistemov SICRIS in COBIS.
S sodelavci smo v zadnjih letih s pomočjo podatkov iz SICRISa uspeli narediti različne analize in primerjave, ki jih bomo predstavili na seminarju.
- Primerjava ekspertnih ocen in bibliometrijskih indikatorjev pri izbiri projektov za financiranje.
- Primerjava raziskovalnih skupin na različnih znanstvenih področjih.
- Mednarodno sodelovanje in vpliv na odmevnost objav.
- Primerjava WoS in SCOPUS.
dr. Janez Povh, Faculty of Information Studies, Novo mesto, Slovenia
From combinatorial optimization to real algebraic geometry and back
Thursday, 16. 10. 2014, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
We will present the most important results in combinatorial optimization and real algebraic geometry and demonstrate how each of these two areas influenced the other area and what is state of the art in both areas.
mag. Andrej Dobrovoljc, the project leader, Faculty of Information Studies, Novo mesto, Slovenia
PhD student Darko Zelenika, collaborator on the project, Faculty of Information Studies, Novo mesto, Slovenia
Undergraduante students Helena Novosel and Robert Pezdirc, collaborators on the project , Faculty of Information Studies, Novo mesto, Slovenia
Document/content capture in small and medium-sized Slovenian enterprises
Friday, 13. 6. 2014, at 11.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
Mikrografija d.o.o. is a Slovenian company, which offers modern solutions for electronic document management and electronic archiving. Their general activity is to capture, process and store paper documents in an electronic form. The problem that Mikrografija d.o.o. faces is that small and medium-sized enterprises have a hard time in adopting automatic data capture software solutions which would convert their paper documents into electronic ones. The main obstacle is the price of such software. Therefore, they decided to offer a simple and easy to use invoice recognition software solution tailored to the enterprises from the Adriatic region, which is reasonably priced. Research and development of such software solution was offered as a project to the Faculty of information studies.
dr. Tadej Kanduč, Faculty of Information Studies, Novo mesto, Slovenia
Optimising production processes in Podgorje
Wednesday, 4. 6. 2014, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
Podgorje d.o.o. is a company that develops, manufactures and sells furniture for caravaning, yachting and special furniture designs (http://www.podgorje.si/).
Within the framework of the project "Creative core: Simulations' our Faculty of information studies is helping Podgorje to find the best layout of machines in order to remove bottlenecks and reduce unnecessary costs. Production system consists of more than one hundred machines and around 30,000 different products. Due to the complexity of the system it is necessary to construct an appropriate simulation model.
On the seminar manufacturing processes will be shown in more detail, a simulation model in Anylogic will be presented and current optimisation results will be revealed.
Co-authorship networks of slovenian scholars: Theoretical analysis and visualization user interface development
Wednesday, 26. 3. 2014, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
The lecturers will present the results of the applied research project: Co-authorship networks of slovenian scholars: Theoretical analysis and visualization user interface development.
The project is funded by the ARRS and T Media d.o.o. and is carried out by Institute of Mathematics, Physics and Mechanics, Jozef Stefan Institute and Faculty of Information Studies in Novo mesto.
dr. Matej Mertik, Faculty of Information Studies, Novo mesto, Slovenia
Mobile application for local food market
Wednesday, 19. 2. 2014, at 11.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
A Pilot Programme of local food market mobile services LOKeT is designed to support an alternative bottom-up approach for strengthening sustainable agricultural practices and local food production. In this presentation Matej Mertik, project leader of the LOKeT project, will present an innovative approach and a mobile service, easily connecting producers and consumers through the open platform LOKeT developed as a collaboration between academia and local industry. The presentation will present LOKeT's open source platform, its functionalities and state of introduction at Centre for agriculture and tourism Grm Novo mesto with future development of the platform undergoing at the Faculty of information studies in Novo mesto.
dr. Aleš Lapajne, Institute of Microbial Sciences and Technologies, Domžale, Slovenia
What is the critical step that follows the genomic and metagenomic analysis of microorganisms?
Wednesday, 8. 1. 2014, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
The lecturer will talk about the following topics:
1. The problems that scientists are deling with in the field of microbiology and presentation of analytical analyses, which provide insight into microbial world.
2. Data structure for 1 degree of metagenomic analysis.
3. Structure of genomic data analysis structure of databases.
4. Data structure in the analyses of the secondary structures of DNA/RNA molecules.
Darko Zelenika, PhD student, Faculty of Information Studies, Novo mesto, Slovenia
Automatic segmentation and categorization of graphically represented text documents
Wednesday, 20. 11. 2013, at 14.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
It is hard to imagine an organization, which does not use computers or the internet in their everyday tasks – to store data, exchange documents, search documents, analyze data, etc. Even though nowadays computers and the internet are a part of practically every organization, and their documents mostly originate on the computer in digital form, the paper documents are still an integral component of their everyday tasks and are likely to remain so for the foreseeable future. Therefore, solutions for automatic document management (recognition and categorization) need to deal with the flow of digital and paper documents in an effective and integrated way. Author will present a disposition of his doctoral dissertation, which aim is to: develop a document segmentation algorithm for the detection (identification) of individual regions in image documents; develop a document categorization algorithm that will be able to take into account information obtained from all the relevant regions of a document; optimize the obtained categorization algorithm for two languages: Slovenian and Croatian. Author will also present the latest results (application mScan) and further work of still ongoing invoice recognition project.
mag. Simona Pustavrh, School center Novo mesto, Novo mesto, Slovenia
Možnost uporabe Geogebre pri izvajanju študijskih programov
Monday, 30. 9. 2013, at 14.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
Uporaba e-storitev, e-gradiv in različnih aplikacij je na vseh stopnjah izobraževanja čedalje pogostejša. Avtorica bo na predavanju predstavila te novosti v srednjih šolah in se osredotočila na uporabo programa Geogebra, ki je na spletu na voljo brezplačno. Prikazala bo možnost uporabe tega programa za boljšo vizualizacijo nekaterih študijskih vsebin, predvsem na primerih za animacijo parametrov normalne porazdelitve in linearnega programiranja.
dr. Panče Panov, Jožef Stefan Institut, Ljubljana, Slovenia
A Modular Ontology of Data Mining
Wednesday, 4. 9. 2013, at 14.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
In this talk, he will present a modular ontology for the domain of data mining named OntoDM. This work was directly motivated by the need for formalization of the data mining domain. The OntoDM ontology is designed and implemented by following ontology best practices and design principles. Its distinguishing feature is that it uses Basic Formal Ontology (BFO) as an upper-level ontology and a template, a set of formally defined relations from the Relational Ontology (RO), and reuses classes and relations from the Ontology of Biomedical Investigations (OBI), the Information Artifact Ontology (IAO), and the Software Ontology (SWO). The OntoDM ontology is composed of three modules covering different aspects of data mining. OntoDT module supports the representation of knowledge about datatypes and is based on an accepted ISO standard for datatypes in computer systems. OntoDM-core module formalizes the key data mining entities for representing the mining of structured data in the context of a general framework for data mining. OntoDM-KDD module formalizes the knowledge discovery process based on the Cross Industry Standard Process for Data Mining (CRISP-DM) process model. The OntoDM ontology was evaluated by assessing the ontology against a set of design principles and best practices, and assessed whether the competency questions posed in the design phase were implemented in the language of the ontology. The developed ontology supports a large variety of applications, such as: the annotation of data mining algorithms; the representation of data mining scenarios; for the annotation of data mining investigations; in cross domain applications to support ontology-based representation of QSAR modeling for drug discovery, as a mid-level ontology by the Expose ontology; and for the annotation of articles containing data mining terms in combination with text mining tools.
dr. Grzegorz Majewski, Faculty of Information Studies, Novo mesto, Slovenia
Deljenje znanja v organizacijah z veliko znanja (primer finskega laboratorija Laurea Lab)
Wednesday, 22. 5. 2013, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
V predavanju bodo predstavljeni rezultati študije, narejene na primeru finskega laboratorija Laurea Lab, in iz nje izpeljan kvantitativni model za deljenje znanja.
Igor Trpevski, Macedonian Academy of Sciences and Arts, Skopje, Macedonia
Napovedovanje z neidealnimi modeli - koncept supermodeliranja
Thursday, 16. 5. 2013, at 13.45, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
Supermodeling by combining imperfect models is a radically different computational strategy that combines ideas from the machine learning, dynamical systems and climate science community to improve our ability to simulate the observed, historical evolution of earth’s climate and obtain more realistic estimates of future climate change. Within the supermodeling concept we take the hypothesis that that one can combine the advantages of different imperfect models to produce an improved forecast of the truth. The combination of the models is determined by a connection mechanism, and the particular values of the connection coefficients are determined using machine learning techniques.
Thus far the usefulness of this concept in improving short-term forecasts over individual models has been demonstrated on the Lorenz 63, the Lorenz 96 family of models, and a quasieostrophic model of the atmosphere. We aim to answer the following set of questions. What assumptions should the imperfect models fulfill to be the suitable components of a supermodel?
What are the main assumptions about the models ensemble - e.g. its completeness, diversity, dimensionality - to be successful in constructing a fully operable and efficient supermodel? A prototypical supermodel that will greatly reduce uncertainty regarding the details of expected climate change, including the magnitude of global warming and specific regional effects. The increased confidence in the projections will facilitate policy decisions at all levels and will increase public support for such decisions, as the public becomes informed about expected changes in each locality, and not just global averages. It has been pointed out that reduction in uncertainty about climate change will itself reduce the cost to society enormously.
Miroslav Mirchev, Ss. Cyril and Methodius University, Skopje, Macedonia
Ocenjevanje stanj in parametrov v Lorenz 96 sistemu
Thursday, 16. 5. 2013, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
We summarize a study of the problem of state and parameter estimation with imperfect models using the Lorenz 96 model that was developed to represent the change in time and space of a certain scalar quantity in the atmosphere. It is assumed that we have a high resolution system referred as a perfect model, with which we generate data. Because practical models are approximations to real processes for modeling we use an imperfect model with lower resolution. Furthermore, in the perfect model we add dynamical noise and in the data we add observational noise. There is only a partial observability of the perfect model and the goal is to estimate the unknown model parameters and unobserved variables to be able to make future predictions.
Two approaches are considered for addressing this problem and in the first approach an optimization problem is set in which we need to find a set of parameters and state variables minimizing a cost function. This approach provide us a maximum-likelihood estimation, but we can not determine the uncertainty of the estimation. In another approach the problem of state and parameter estimation can be set as a path integral over the states conditioned on the observed data and this approach further give us the statistics of the estimation.
dr. Kristina Šorić, Zagreb School of Economics and Management, Zagreb, Croatia
Metahevristike za dvokriterijalno optimizacijo vodenja živilskih oskrbovalnih verig
(Metaheuristics for Bicriteria Optimization in Agri-Food Supply Chain Management)
Wednesday, 17. 4. 2013, at 11.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
The seminar addresses a problem of supply chain management in olive oil industry inspired by a real case from Croatia. The focus is on scheduling three phases of considered supply chain, olives harvesting and delivering performed by suppliers and olive oil production performed by producers. The suppliers want to minimize the harvest delay and delivery cost while the producers maximize the profit consisting of revenue, production cost and storage cost. We model this problem as a bi-criteria mixed integer programming problem considering the suppliers as the first group of decision makers and the producers as the second group of decision makers. The conflict in their objectives is in the fact that suppliers want to deliver their olives immediately upon the harvesting to keep the quality of the oil high while the producers want to avoid the storage of olives if there is no available production capacity in the moment of delivery. Due to the big dimension of the problem two metaheuristics are considered, scatter search based method (SS) and non-dominated sorting genetic algorithm (NSGA-II). At the end some simulations are performed and two metaheuristics compared.
Vesna Andova, PhD student, Faculty of Electrical Engineering and Information Technology, Ss. Cyril and Methodius University, Skopje, Macedonia
Izbrani rezultati o fullerenskih grafih
Wednesday, 10. 4. 2013, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
Fullerenski grafi so 3-povezani 3-regularni ravninski grafi, ki imajo le pentagonalna in heksagonalna lica. Uporabljamo jih za opisovanje strukture karbonskih molekul. Posebej zanimiv razred teh grafov so nanocevi zaradi uporabe pri razvoju različnih nanotehnologij. Zaradi širokega spektra možnih uporab so ti grafi deležni velike pozornosti širokega spektra raziskovalnih skupnosti, od katerih se mnoge ukvarjajo s problemom identifikacije strukturnih lastnosti stabilnih fullerenov. Na seminarju bo podan kratek pregled fullerenskih grafov, nato pa bodo predstavljeni nekateri novi rezultati glede stabilnosti fullerenov.
dr. Blaž Fortuna, Artificial Intelligence Laboratory, Jožef Stefan Institute, Ljubljana, Slovenia
Real-time news recommendation with rich representation
Wednesday, 20. 3. 2013, at 13.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
News recommendation is an area of research where we deal with a non-stationary source of documents which are recommended to the users of the publishers' web sites. Predominant success metric is the attention span of a user expressed in terms of time spent on site and number page views. The key modeling problem is the fact that the most relevant news to be recommended are usually the fresh ones having no usage history, ie. the goal is to recommend items about which we don't know much.
There are several types of data one consider when doing news recommendation. The most obvious ones are content of the articles and collaborative filtering with the help of contextual features like GeoIP, time, and demographics. More sophisticated types of data include semantics extracted from the text, meta data and inferred demographics (look-a-likes). Once having a representation determined, an important dimension is granularity of modeling for personalized information delivery balanced with the required response time (processing speed).
In this contribution we will present a solution using most of the above ingredients built for a large online business news providers with up-to few hundred page views per second. The talk will focus on design decisions leading to a successful self adaptive system serving millions of users per day.
dr. Marko Maras, Agricultural Institute of Slovenia, Ljubljana, Slovenia
Raziskovalne metode in orodja v aplikativni genetiki rastlin
sreda, 20. 2. 2013, ob 13. uri v sejni sobi FIŠ (Ulica talcev 3, Novo mesto)
Med največje dosežke znanosti v 20. stoletju nedvomno sodi odkritje dvoverižne strukture DNA molekule leta 1953. Desetletja, ki so sledila, je zaznamoval intenziven razvoj novih metod in orodij v genetiki. Nov mejnik v disciplini je konec sedemdesetih let prispeval razvoj tehnologije branja zaporedja nukleotidnega zaporedja DNA molekule, kar je kasneje omogočilo identifikacijo genov, opredeljevanje njihove funkcije v organizmih ter njihove sinteze v laboratoriju. Nova odkritja in spoznanja so bila povod za nastanek novih disciplin, kot so molekulska biologija, strukturna in funkcijska genomika, transkriptomika, proteomika, bioinformatika, sistemska biologija idr. Temeljna spoznanja iz vseh naštetih ved so se do danes vsaj delno že prelila v prakso tako v humani medicini kot tudi v animalni in rastlinski produkciji. V referatu se bomo seznanili z nekaterimi splošnimi orodji in metodami, ki jih danes na Kmetijskem inštitutu Slovenije uporabljamo v genetskih raziskavah na rastlinah.
dr. Rok Košir, DiaGenomi d.o.o., Ljubljana, Slovenia
S patologijo podprta genetska testiranja
sreda, 9. 1. 2013, ob 13. uri v sejni sobi FIŠ (Ulica talcev 3, Novo mesto)
S patologijo podprta genetska testiranja (angl. Pathology supported genetic testing - PSGT) predstavljajo nov koncept v personalizirani medicini. V nasprotju z vedno bolj priljubljenimi DTC genetskimi testi (angl. direct to consumer genetic testing), ki preverijo samo vašo DNA, PSGT testi zahtevajo tudi natančne odgovore na vprašanja o življenskem stilu, vključujoč prehranjevalne navade, rekreacijo, uzivanje alkohola itd., zdravstveno anamnezo in krvno sliko. Integracija vseh teh informacij nam omogoča pripravo personalizirane ocene tveganja in natančna navodila za spemembo življenjskega stila z namenom, da se izognemo razvoju bolezni.
dr. Nataša Pržulj, Imperial College, London
Network topology as a source of information
Wednesday, 7. 11. 2012, at 14.00, seminar room of the Faculty on information studies (Ulica talcev 3, Novo mesto, Slovenia)
Many real-world phenomena can be represented as networks of interconnected entities. For example, individual genes are just a means to an end: they produce proteins that interact in complex networked ways and make our cells work. Hence, using protein interaction networks (PINs) to predict protein function and involvement in disease has received much attention in the post-genomic era. We develop novel measures of network topology to predict function of unannotated proteins in the human PIN. We find that human genes involved in key biological processes and pathways, such as aging, cancer, infectious diseases, signaling and drug-targeted pathways, occupy regions of the network that correspond to its "spine" that connects all other network parts and can thus pass cellular signals fast throughout the network. We design methods that harvest information from network topology and gain new biological information, such as suggest novel drug targets for therapeutic intervention. For example, our network-based predictions of novel proteins that participate in melanogenesis in human cells are phenotypically validated.