Ciclo de Seminarios MAVIR

lunes 24 - miércoles 26 de noviembre de 2008, UC3M y UNED

Andreas Nürnberger: "Adaptativity and Context in Information Retrieval Systems"

Marcin Detyniecki: "Multimedia Retrieval and the Semantic Gap"

Paul Clough: "Key Technologies for Business Intelligence" and "Authorship Attribution and Plagiarism Detection"

Adaptativity and Context in Information Retrieval Systems

Ponente: Prof. Dr. Andreas Nürnberger (Otto-von-Guericke-Universität Magdeburg)
Horario: lunes 24/11 y martes 25/11, dos sesiones por la mañana 10h-14h

Lugar de celebración presencial:

Lugar de celebración por videoconferencia:

Resumen:

Currently, a huge amount of methods and tools is available that tries to tackle the problem of searching and organizing multimedia collections. The approaches range from simple category based methods over social plattforms like flickr for images of last.fm for music to tools that try to make use of content based features in order to enable semantic searches.

This course gives a quick introduction to the underlying methods and ideas. Furthermore, recent developments and research issues are discussed. We will start with an introduction of fundamental concepts of text and multimedia retrieval systems, i.e. methods for indexing, feature extraction, preprocessing and appropriate data structures. Then, concepts for retrieval systems will be discussed that enable the use of user specific information to search, structure, visualize and organize information archives. Besides the algorithms itself also aspects of user interface design will be considered.

Multimedia Retrieval and the Semantic Gap

Ponente: Dr. Marcin Detyniecki (CNRS & Carnegie Mellon University)

Horario: lunes 24/11 y martes 25/11, dos sesiones por la tarde 16h-20h

Lugar de celebración:

Resumen:

This lecture will present how a multimedia search engine works and what are the fundamental differences with the text engines. To understand how such an system works we will focus on four points: data, indexation, querying and visualization. In the data part, we will have a look at what multimedia data is about and what are the new challenges that it naturally brings with. In the query part, we will exhibit a number of new types of queries and interactions (other than text keywords) that are possible with multimedia data. In the visualization part we discuss some of the new problems appearing when retrieving non textual documents. In the indexation part we see what can be today extracted from an image, from a sound, from a video. We will see that, what is extracted is far away from the meaning given by a user when looking at the multimedia document. This mismatch is known as the semantic gap. In the last part of the course, and by far the longest one, we will present how this gap can be reduced (without human intervention). 

Key Technologies for Business Intelligence 

Ponente: Dr. Paul Clough (University of Sheffield)

Horario: miércoles 26/11, 10h-14h

Lugar de celebración:

Resumen:

In the modern business environment, organisations face multiple challenges in maintaining their commercial edge. Business Intelligence (BI) is one of many tools an organisation can use to improve their business decision-making processes. It focuses on identifying emerging issues, situations and potential pitfalls that may influence the future of an organisation. Although not a new concept in business, an increase in the availability of digital content has changed (or could change) how many organisations conduct BI activities.

In this lecture I will introduce some of the key technologies commonly used to support BI, including data management, data mining, visualisation and the use of natural language technologies. I will begin the lecture by firstly describing the operational context (i.e. the business environment), the notion of BI within the organisation, and then introduce a number of key technologies aimed at supporting BI processes. This lecture provides an introduction to the topic, and therefore does not require any subject-specific knowledge and would be suitable for both technical and non-technical students.

Authorship Attribution and Plagiarism Detection

Ponente: Dr. Paul Clough (University of Sheffield)

Horario: miércoles 26/11, 16h-20h

Lugar de celebración:

Resumen:

Often the focus in retrieval of written texts is on finding documents which are thematically or topically relevant to the user’s query. However, there are other “dimensions” from which texts can be analysed. For example, style, genre and register, authorship, difficulty or readability, and novelty. Typically computational linguists study these areas which not only have real-world applications, but are complementary to the notions of traditional Information Retrieval (IR).

In this lecture, I will present two specific text analysis problems: authorship attribution and plagiarism detection. The lecture will introduce these topics and demonstrate how both authorship style and plagiarism can be detected. This lecture provides an introduction to the topic, and therefore does not require any subject-specific knowledge and would be suitable for both technical and non-technical students. 

Materiales

Próximamente...