Student Corner

Master Thesis Offerings

Begin: as soon as possible (January or February 2012).

Duration: 6-8 months.

IF YOU HAVE AN IDEA CONCERNING A THESIS (Master, Bachelor) THAT IS RELEVANT TO OUR RESEARCH, PLEASE CONTACT US!

Concrete Master Thesis Offerings

Title: A News Retrieval Engine for financial search queries

The goal of the Master thesis is to work with Reuters Financial News data and to design and implement a news retrieval engine on the basis of given search term sequences. In this work, we may assume that only search terms, which are spelled correctly, are considered. Example: the user enters 'Eurobond Juncker' and the search engine delivers all news documents where the search term is in at least once. For this, the existing terms in the documents must be indexed first; second, postings lists must be established. The search engine should be able to reference its individual results to a document number ID including a document relevance score (how good does a search query fits to the financial document.

This Master Thesis is related to the research project ESCAPE, which is funded by the FNR (CORE program) and which will start in 2012 for a duration of 3 years. ESCAPE is a common project of the Luxembourg School of Finance (LSF, Prof. Grammatikos) and the Computer Science Department (CSC, Prof. Schommer). ESCAPE applies Machine Learning methods on financial news articles (Thomson Reuters, New York Times) to document in a measurable way the structure and evolution of Europe's financial policy and to address the on-going threat to the Eurozone stability.

Contact: Prof. Schommer, Email: christoph.schommer @ uni.lu

String matching-algorithms for search queries in Financial Texts

The problem of how to deal with misspellings in user-based search terms is the concern of this master thesis. How can invalid terms be transformed to valid terms? What will be the most likely substitutes? How expensive is such a transformation? There exist a variety of solutions to this problem, for example n-grams, Levenshtein, Robin-Karp, Knuth-Morris, et cetera. The major content of the thesis is the implementation of such algorithms, the establishment of an human-computer interface, and the solid testing against the financial news databases we have (Thomson Reuters, New York Times). Keyword: Information Retrieval, String-matching, Financial News Articles.

This Master Thesis is related to the research project ESCAPE, which is funded by the FNR (CORE program) and which will start in 2012 for a duration of 3 years. ESCAPE is a common project of the Luxembourg School of Finance (LSF, Prof. Grammatikos) and the Computer Science Department (CSC, Prof. Schommer). ESCAPE applies Machine Learning methods on financial news articles (Thomson Reuters, New York Times) to document in a measurable way the structure and evolution of Europe's financial policy and to address the on-going threat to the Eurozone stability.

Contact: Prof. Schommer, Email: christoph.schommer @ uni.lu

Wild-card search in Financial News Textss

The problem of how to deal with wild-card searches in user-based search terms (Example: "Eurob*", "Jun*er") is the concern of this master thesis. Generally, a wild-card search is rather expensive, since all potential terms in the vocabulary must be presented to the user. But which of them make sense and which not? Can we learn from the user's behavior and favour some terms more than others? How to deal with wild-cards that occur in the middle of a string? Established techniques implicate for example the usage n-grams, permuterm indexes, B-trees, et cetera. The goal of the thesis is to implement some these techniques, to establish a human-machine interface, and to test the algorithms against the financial news databases we have (Thomson Reuters, New York Times). Keyword: Information Retrieval, Wild-card Search, Financial News Articles.

This Master Thesis is related to the research project ESCAPE, which is funded by the FNR (CORE program) and which will start in 2012 for a duration of 3 years. ESCAPE is a common project of the Luxembourg School of Finance (LSF, Prof. Grammatikos) and the Computer Science Department (CSC, Prof. Schommer). ESCAPE applies Machine Learning methods on financial news articles (Thomson Reuters, New York Times) to document in a measurable way the structure and evolution of Europe's financial policy and to address the on-going threat to the Eurozone stability.

Contact: Prof. Schommer, Email: christoph.schommer @ uni.lu

Title: Topic Detection of Financial Texts

Financial News articles contain different content, for example the time, the place, names of persons, thematic issue and relevance, and others. Often, this content is thematically not explicitly separated, but kept as a mixture. The goal of this master thesis is to define and identify topics in the financial news domain and to assign these to a thematic zones. If done, what can we say about the similarity of financial news articles? Keywords: Topic Detection, Zoning, Similarity, Clustering, Financial News.

This Master Thesis is related to the research project ESCAPE, which is funded by the FNR (CORE program) and which will start in 2012 for a duration of 3 years. ESCAPE is a common project of the Luxembourg School of Finance (LSF, Prof. Grammatikos) and the Computer Science Department (CSC, Prof. Schommer). ESCAPE applies Machine Learning methods on financial news articles (Thomson Reuters, New York Times) to document in a measurable way the structure and evolution of Europe's financial policy and to address the on-going threat to the Eurozone stability.

Contact: Prof. Schommer, Email: christoph.schommer @ uni.lu

Title: Information Hiding in Financial Texts

The Master Thesis is to concern with the field of linguistic steganography, which refers to a conscious hiding of a secret textual message inside a text. Many approaches exist, for example random and statistical approaches, text mimicking, or diverses contributions from semantics. The goal of the master thesis is to design and implement several of these strategies and to apply these techniques for financial news texts. A evaluation of the steg-engine must be performed: how 'good' is the encryption with respect to original sources? Keywords: Linguistic Steganography, information Hiding, Financial News Articles.

This Master Thesis is related to 2 research projects: a) SALT, which starts in January 2012 (duration 2 years) and which deals with Linguistic Steganography and the role of mimicry, and b) to the research project ESCAPE, which is funded by the FNR (CORE program) and which will start in 2012 for a duration of 3 years. ESCAPE is a common project of the Luxembourg School of Finance (LSF, Prof. Grammatikos) and the Computer Science Department (CSC, Prof. Schommer). ESCAPE applies Machine Learning methods on financial news articles (Thomson Reuters, New York Times) to document in a measurable way the structure and evolution of Europe's financial policy and to address the on-going threat to the Eurozone stability.

Contact: Prof. Schommer, Email: christoph.schommer @ uni.lu

Title: Link Analysis in Financial News

Link analysis represents a class of techniques that are used to evaluate relationships between nodes in a graph. Among others, relationships might exist among various types of persons and organizations. The goal of the master thesis is to link persons and institutions within financial news and to monitor their relationship over time. Keywords: Link Analysis, Financial News.

This Master Thesis is related to the research project ESCAPE, which is funded by the FNR (CORE program) and which will start in 2012 for a duration of 3 years. ESCAPE is a common project of the Luxembourg School of Finance (LSF, Prof. Grammatikos) and the Computer Science Department (CSC, Prof. Schommer). ESCAPE applies Machine Learning methods on financial news articles (Thomson Reuters, New York Times) to document in a measurable way the structure and evolution of Europe's financial policy and to address the on-going threat to the Eurozone stability.

Contact: Prof. Schommer, Email: christoph.schommer @ uni.lu

Addendum

Some thoughts

Some thoughts about How to write a thesis (pdf) in the field of Computer Science (to be continued).
Some thoughts about How to write a presentation - the P5 strategy (pdf) in the field of Computer Science (to be continued).

Submitting a Research Proposal

Doctoral, Master, and Bachelor students are encouraged to propose an own working proposal in the fields of

Data Mining and Knowledge Discovery.
Machine Learning and Information Extraction.
Text Mining and Document Analysis.
Information Search and Retrieval.
Self-organisation and Self-adaption.

Depending on the academic level, the attitude to work, the performance, and the conditions given in the program of study, the duration of a thesis varies from several months (Bachelor, Master) to 3-4 years (doctorate).