Masoud Makrehchi
PhD
Associate Professor
Electrical, Computer and Software EngineeringFaculty of Engineering and Applied Science
Leading social media research analytics to predict trends in behaviour for safer communities
masoud.makrehchi@ontariotechu.ca
905.721.8668 ext. 5387
- PhD - Electrical and Computer Engineering University of Waterloo 2007
- MSc - Computer Engineering Shiraz University 1994
- BSc - Software Engineering Iran University of Science and Technology 1991
How to Predict Social Trends by Mining User Sentiments
Washington, D.C. March 31, 2015International Conference on Social Computing, Behavioral-Cultural Modeling, and Prediction (SBP-2015)
The Correlation Between Language Shift and Social Conflicts in Polarized Social Media
Warsaw, Poland August 11, 2014IEEE/WIC/ACM International Conference on Web Intelligence
Stock Prediction Using Event-based Sentiment Analysis
Atlanta, Georgia November 17, 2013IEEE/WIC/ACM International Conference on Web Intelligence
Social Links Recommendation by Learning Hidden Topics
Chicago, Illinois January 1, 2011Association for Computing Machinery (ACM) Conference on Recommender Systems
Query-relevant Document Representation for Text Clustering
Thunder bay, Ontario July 5, 2010Fifth International Conference on Digital Information Management
Text Clustering Using Taxonomy
Sydney, Australia December 9, 20082008 IEEE/WIC/ACM International Conference on Web Intelligence
Automatic Extraction of Domain-Specific Stopwords from Labeled Documents
Glasgow, Scotland March 30, 20082008 European Conference on Information Retrieval
A Text Classification Framework with a Local Feature Ranking for Learning Social Networks
Omaha, Nebraska October 28, 20072007 IEEE International Conference on Data Mining
Evaluating Feature Ranking Methods in Text Classifiers
Published in Intelligent Data Analysis September 1, 2015Masoud Makrehchi
In this paper, a framework, which is called feature meta-ranking, is introduced to identify the best feature ranking measure among a set of candidate solutions for a particular text classification problem. The feature meta-ranking technique is implemented based on the differential filter level performance method. This method uses a simple classifier, such as Rocchio, to estimate the behaviour of the feature ranking measure with respect to a particular data set. With respect to the use of a classifier in the feature selection loop, the proposed method can be considered as a hybrid feature selection technique with minimal use of a classifier in the loop. The proposed method is evaluated by applying it to six data sets.
View more - Evaluating Feature Ranking Methods in Text Classifiers
The Correlation Between Language Shift and Social Conflicts in Polarized Social Media
Published in Web Intelligence and Intelligent Agent Technologies August 11, 2014Masoud Makrehchi
In a polarized society, rhetorical arguments are usually expressed by strong, extreme terms which by themselves carry a positive or negative sentiment about one side of the social debate or conflict. By detecting extreme terms in a social-political text such as a blog post, we are able to automatically detect the sentiment of the text about polarizing issues in a divided society. On the other hand, during social and political conflicts in polarized societies, we observe a shift from mainstream to extreme language and rhetoric. In this research, we illustrate that there is a correlation between language shift and social conflicts. In other words, the language shift can be used as a signal for predicting social conflicts in divided societies. The data for this research were collected from Iranian political blogs during the Iran election crisis in 2009.
View more - The Correlation Between Language Shift and Social Conflicts in Polarized Social Media
Feature Ranking Fusion for Text Classifiers
Published in Intelligent Data Analysis November 1, 2012Masoud Makrehchi & Mohamed S. Kamel
Feature ranking is widely used in text classification. One problem with feature ranking methods is their non-robust behaviour when applied to different data sets. In other words, the feature ranking methods behave differently from one data set to the other. The problem becomes more complex when we consider that the performance of feature ranking methods highly depends on the type of text classifier. In this paper, a new method based on combining feature rankings is proposed to find the best features among a set of feature rankings. The proposed method is applied to the text classification problem and evaluated on three well-known data sets using Support Vector Machine and Rocchio classifier. Several combining methods are employed to aggregate ranked lists of features. We show that combining methods can offer reliable results very close to the best solution without the need to use a classifier.
An Information Theoretic Approach to Generating Fuzzy Hypercubes for If-Then Classifiers
Published in Journal of Intelligent and Fuzzy Systems January 1, 2011Masoud Makrehchi & Mohamed S. Kamel
In this paper, a framework for automatic generation of fuzzy membership functions and fuzzy rules from training data is proposed. The main focus of this paper is designing fuzzy if-then classifiers; however, the proposed method can be employed in designing a wide range of fuzzy system applications. After the fuzzy membership functions are modelled by their supports, an optimization technique, based on a multi-objective real coded genetic algorithm with adaptive cross over and mutation probabilities, is implemented to find near-optimal supports.
View more - An Information Theoretic Approach to Generating Fuzzy Hypercubes for If-Then Classifiers
Impact of Term Dependency and Class Imbalance on the Performance of Feature Ranking Methods for Text Classifiers
Published in International Journal of Pattern Recognition and Artificial Intelligence November 1, 2011Masoud Makrehchi and Mohamed S. Kamel
Feature ranking is widely employed to deal with high dimensionality in text classification. The main advantage of feature ranking methods is their low cost and simple algorithms. However, they suffer from some drawbacks which cause low performance compared to wrapper approach feature selection methods.
Taxonomy-based Document Clustering
Published in Journal of Digital Information Management April 1, 2011Masoud Makrehchi
One well-known document representation for text clustering is bag-of-words. Although it is simple and popular, it ignores semantics, underlying linguistic information, and word correlations. In this paper, Bag-Of-Queries, a new document representation is proposed.
Filter-based Data Partitioning for Training Multiple Classifier Systems
Published in IEEE Transactions on Knowledge and Data Engineering February 1, 2010Rozita A. Dara, Masoud Makrehchi & Mohamed S. Kamel
This study is concerned with the analysis of training data distribution and its impact on the performance of multiple classifier systems. In this study, several feature-based and class-based measures are proposed. These measures can be used to estimate the statistical characteristics of the training partitions. To assess the effectiveness of different types of training partitions, we generated a large number of disjoint training partitions with distinctive distributions. Then, we empirically assessed these training partitions and their impact on the performance of the system by utilizing the proposed feature-based and class-based measures. We applied the findings of this analysis and developed a new partitioning method called "Clustering, De-clustering, and Selection" (CDS). This study presents a comparative analysis of several existing data partitioning methods including our proposed CDS approach.
View more - Filter-based Data Partitioning for Training Multiple Classifier Systems
Best Paper Award
2014 International Conference on Social Computing, Behavioral-Cultural Modeling, and Prediction (SBP14) April 2, 2014Awarded for his co-authored, peer-reviewed, research paper Winning by Following the Winners: Mining the Behavior of Stock Market Experts in Social Media at the SBP14 Conference in Washington, D.C.
Senior Member
IEEE March 1, 2013Dr. Makrehchi has made major contributions to his field, advancing social media mining and analytics techniques to predict behavioural outcomes.
Challenge Award
2012 International Conference on Social Computing, Behavioral-Cultural Modeling, and Prediction (SBP12) April 3, 2012Dr. Makrehchi received the award for his research Conflict Thermometer: Predicting Social Conflicts by Analyzing Language Gap in Polarized Social Media at SBP12 in College Park, Maryland.
Towards Predicting Socio-Economic Systems by Mining Social Media Data
NSERC Discovery Grant April 1, 2014The basis of this five-year research project is to analyze social media data to develop algorithms for predicting trends in crime and public sentiment, and moods related to market sentiment. The research also explores the science behind expressive writing and investigates ways to detect certain neurological disorders from expressive writing. ($75,000)
Social Media, Investor Sentiment and Initial Public Offerings
SSHRC Insight Development Grant February 1, 2014This two-year research project analyzes social media data to develop algorithms for predicting trends in crime and public sentiment, and moods related to market sentiment. ($74,718)
Computer Assisted Generation and Transformation of Web Content
NSERC Engage Grant April 1, 2014In collaboration with Search Engine People, a leading digital marketing firm, this research project focuses on natural language regeneration, and uses a machine learning approach to process naturally found language from structured data.
Institute of Electrical and Electronics Engineers (IEEE)
Professional Engineers Ontario
- Design and Analysis of Algorithms (SOFE 3770U)
Designing and analyzing algorithms; asymptotic notation; recurrences and recursion; probabilistic analysis and randomized algorithms; sort algorithms; priority queues; medians and order statistics; data and advanced data structures; augmenting data structures for custom applications; dynamic programming; greedy algorithms; graph algorithms; sorting networks; matrix operations; linear programming; number-theoretic algorithms; string matching; NP-completeness and approximation algorithms; object libraries. - Introduction to Artificial Intelligence (SOFE 3720U)
This course introduces students to basic concepts and methods of artificial intelligence from a software engineering perspective. The emphasis of the course will be on the selection of data representations and algorithms useful in the design and implementation of intelligent systems. Knowledge representation methods, state-space search strategies, and use of logic for problem-solving. Applications are chosen from among expert systems, planning, natural language understanding, uncertainty reasoning, machine learning, and robotics. The course will contain an overview of one AI language and discussion of important applications of artificial intelligence methodology. - Real-Time Systems and Control (SOFE 4830U)
Computing systems design for real-time applications in control, embedded systems and communications; microcontrollers; data acquisition in robotics and manufacturing, file management, memory management and multitasking in a real-time environment; object-oriented design principles for real-time systems. Robustness - Knowledge Discovery and Data Mining (ENGR 5775G)
This course covers the discovery of new knowledge using various data mining techniques on real-world datasets, and the current research directions represent the foundation context for this course. This course utilizes the latest blended learning techniques to explore topics in foundations of knowledge discovery and data mining; data mining approaches; and the application of data mining within such diverse domains as health care, business, supply chain and IT security. Current research directions, trends, issues and challenges are also explored.