PhD Defence: Sentiment Analysis of Text Guided by Semantics and Structure


In his dissertation ‘Sentiment Analysis of Text Guided by Semantics and Structure, ERIM’s Alexander Hogenboom argues that the polarity of text can be analysed more accurately when additionally accounting for semantics and structure.  As moods and opinions play a pivotal role in various business and economic processes, keeping track of one's stakeholders' sentiment can be of crucial importance to decision makers. Today's abundance of user-generated content allows for the automated monitoring of the opinions of many stakeholders, like consumers. One challenge for such automated sentiment analysis systems is to identify whether pieces of natural language text are positive or negative.

 

 

Alexander Hogenboom will defend his dissertation in the Senate Hall, Erasmus Building at Erasmus University Rotterdam on Friday, 13 November 2015, at 13:30. His supervisors are Prof.dr.ir. U. Kaymak , Prof.dr. F.M.G. de Jong and Dr.ir. F. Frasincar . Other members of the Doctoral Committee are Prof.dr. P.J.F. Groenen (ESE), Prof.dr. M.-F. Moens (KU Leuven), and Dr. D.E. Losada (Universidad de Santiago de Compostela).

 

Alexander Hogenboom’s PhD research project is conducted within the Erasmus Doctoral Programme in Business and Management organised by Erasmus Research Institute of Management (ERIM), the joint research institute of Rotterdam School of Management (RSM) and Erasmus School of Economics (ESE) of the Erasmus University Rotterdam (EUR).

 

About Alexander Hogenboom


Alexander Hogenboom (April 13, 1987) holds both a B.Sc. degree and a cum laude M.Sc. degree in Economics and Informatics, obtained at Erasmus University Rotterdam, The Netherlands, in 2007 and 2009, respectively. His research interests relate to the utilization of tools and techniques stemming from computer science in order to facilitate or support (business) economic processes. As such, Alexander's research covers semantic information systems, decision support systems, and intelligent systems for information extraction, with a specific focus on systems for automated sentiment analysis.

 

Since July 2009, Alexander has conducted his research in a Ph.D. candidacy under the auspices of the Erasmus Center for Business Intelligence (ECBI) at the Erasmus Research Institute of Management (ERIM), the Econometric Institute at the Erasmus School of Economics (ESE), Erasmus Studio, and the Dutch Research School for Information and Knowledge Systems (SIKS). Alexander's Ph.D. research is linked to the Argumentation Discovery in Economics Literature project of ERIM, the Semantic Scholarly Publishing project of Erasmus Studio, and the Infiniti project on Information Retrieval for Information Services (work package three) of the Dutch national program COMMIT.

 

Alexander has published numerous (peer-reviewed) papers in the proceedings of prestigious international conferences – e.g., BIS, CIKM, DEXA, ER, ICEC, NLDB, SAC, SMC, and WISE – and local venues like BNAIC, DBDBD, and DIR. His conference papers brought him an Honorable Mention Award (at ICEC 2009), as well as various travel grants. Moreover, Alexander has published several articles in renowned journals such as Communications of the ACM, Data and Knowledge Engineering, the Decision Sciences Journal, Decision Support Systems, Expert Systems with Applications, and the Journal of Web Engineering. Alexander is also a contributor to the EconomieOpinie and Backbone platforms.

 

In addition to his research activities, Alexander has acted as a reviewer for renowned journals like the Expert Systems with Applications journal and the Information Systems journal. Furthermore, he has been actively involved with international conferences, not only as participant, but also as a local organizer, session chair, program committee member, and reviewer.

 

Over the years, Alexander has played an active role at the Erasmus University Rotterdam as well. Between October 2009 and May 2013, he was a board member of the Erasmus Ph.D. Association Rotterdam (EPAR). In this position, he was responsible for internal and external communication. Furthermore, since July 2009, Alexander has been intensively involved with the supervision of various Bachelor's and Master's theses. Additionally, he has been involved with coordinating and teaching many courses related to computer programming and information technology. Alexander's teaching activities have resulted in excellent student reviews, culminating in a nomination for a Professor of the Year Award for the first year of the International Business Administration (IBA) Bachelor's programme at the Rotterdam School of Management (RSM) in 2014.

 

 

Thesis Abstract


As moods and opinions play a pivotal role in various business and economic processes, keeping track of one's stakeholders' sentiment can be of crucial importance to decision makers. Today's abundance of user-generated content allows for the automated monitoring of the opinions of many stakeholders, like consumers. One challenge for such automated sentiment analysis systems is to identify whether pieces of natural language text are positive or negative.

 

Typical methods of identifying this polarity involve low-level linguistic analysis. Existing systems predominantly use morphological, lexical, and syntactic cues for polarity, like a text's words, their parts-of-speech, and negation or amplification of the conveyed sentiment. This dissertation argues that the polarity of text can be analysed more accurately when additionally accounting for semantics and structure.

 

Polarity classification performance can benefit from exploiting the interactions that emoticons have on a semantic level with words – emoticons can express, stress, or disambiguate sentiment. Furthermore, semantic relations between and within languages can help identify meaningful cues for sentiment in multi-lingual polarity classification.

 

An even better understanding of a text's conveyed sentiment can be obtained by guiding automated sentiment analysis by the rhetorical structure of the text, or at least of its most sentiment-carrying segments. Thus, the sentiment in, e.g., conclusions can be treated differently from the sentiment in background information.

 

The findings of this dissertation suggest that the polarity of natural language text should not be determined solely based on what is said. Instead, one should account for how this message is conveyed as well.