Bahadorreza Ofoghi

Data Mining in Elite Sports: A Review and a Framework

Sophisticated data analytical methods such as data mining, where the focus is upon exploration and developing new insights, are becoming increasingly useful tools in analysing elite sports performance data and supporting decision making that is crucial to gaining success. In this article, we investigate the different data mining demands of elite sports with respect to a number of features that describe sport competitions. The aim is to more structurally connect the sports and data mining domains through: (a) describing a framework for categorizing elite sports, and (b) understanding the analytical demands of different performance analysis problems. Therefore, we review different aspects such as sport categories and performance analysis requirements that influence each stage in sports data mining. We also present a model bringing together performance analysis requirements, data mining methods, data mining techniques, and technique characteristics. This will assist both data scientists and sport professionals to more effectively collaborate and contribute to success in elite sport events.

TOWARDS EARLY DISCOVERY OF SALIENT HEALTH THREATS: A SOCIAL MEDIA EMOTION CLASSIFICATION TECHNIQUE

Online social media microblogs may be a valuable resource for timely identification of critical ad hoc health-related incidents or serious epidemic outbreaks. In this paper, we explore emotion classification of Twitter microblogs related to localized public health threats, and study whether the public mood can be effectively utilized in early discovery or alarming of such events. We analyse user tweets around recent incidents of Ebola, finding differences in the expression of emotions in tweets posted prior to and after the incidents have emerged. We also analyse differences in the nature of the tweets in the immediately affected area as compared to areas remote to the events. The results of this analysis suggest that emotions in social media microblogging data (from Twitter in particular) may be utilized effectively as a source of evidence for disease outbreak detection and monitoring.

A hybrid question answering schema using encapsulated semantics in lexical resources

Question Answering systems, as an extreme of the Information Retrieval field, could save lots of time and effort in satisfying a specific information need. In this regard, there are still many challenges to be resolved by current state-of-the-art systems as they cope with free texts. We propose a new hybrid question answering schema capable of answering questions with respect to different semantically related syntactically mismatched situations either in a structured or unstructured semantic format. We have exploited FrameNet and WordNet lexical resources and implemented the prototype system in a TREC-friendly fashion to obtain results comparable with outstanding participant systems in TREC 2004.

A machine learning approach to predicting winning patterns in track cycling omnium

This paper presents work on using Machine Learning approaches for predicting performance patterns of medalists in Track Cycling Omnium championships. The omnium is a newly introduced track cycling competition to be included in the London 2012 Olympic Games. It involves six individual events and, therefore, requires strategic planning for riders and coaches to achieve the best overall standing in terms of the ranking, speed, and time in each individual component. We carried out unsupervised, supervised, and statistical analyses on the men’s and women’s historical competition data in the World Championships since 2008 to find winning patterns for each gender in terms of the ranking of riders in each individual event. Our results demonstrate that both sprint and endurance capacities are required for both men and women to win a medal in the omnium. Sprint ability is shown to have slightly more influence in deciding the medalists of the omnium competitions.

Modelling and analysing track cycling Omnium performances using statistical and machine learning techniques

Abstract This article describes the utilisation of an unsupervised machine learning technique and statistical approaches (e.g., the Kolmogorov-Smirnov test) that assist cycling experts in the crucial decision-making processes for athlete selection, training, and strategic planning in the track cycling Omnium. The Omnium is a multi-event competition that will be included in the summer Olympic Games for the first time in 2012. Presently, selectors and cycling coaches make decisions based on experience and intuition. They rarely have access to objective data. We analysed both the old five-event (first raced internationally in 2007) and new six-event (first raced internationally in 2011) Omniums and found that the addition of the elimination race component to the Omnium has, contrary to expectations, not favoured track endurance riders. We analysed the Omnium data and also determined the inter-relationships between different individual events as well as between those events and the final standings of riders. In further analysis, we found that there is no maximum ranking (poorest performance) in each individual event that riders can afford whilst still winning a medal. We also found the required times for riders to finish the timed components that are necessary for medal winning. The results of this study consider the scoring system of the Omnium and inform decision-making toward successful participation in future major Omnium competitions.

Learning Parse-Free Event-Based Features for Textual Entailment Recognition

We propose new parse-free event-based features to be used in conjunction with lexical, syntactic, and semantic features of texts and hypotheses for Machine Learning-based Recognizing Textual Entailment. Our new similarity features are extracted without using shallow semantic parsers, but still lexical and compositional semantics are not left out. Our experimental results demonstrate that these features can improve the effectiveness of the identification of entailment and no-entailment relationships.

Performance analysis and prediction in triathlon

Abstract Performance in triathlon is dependent upon factors that include somatotype, physiological capacity, technical proficiency and race strategy. Given the multidisciplinary nature of triathlon and the interaction between each of the three race components, the identification of target split times that can be used to inform the design of training plans and race pacing strategies is a complex task. The present study uses machine learning techniques to analyse a large database of performances in Olympic distance triathlons (2008–2012). The analysis reveals patterns of performance in five components of triathlon (three race “legs” and two transitions) and the complex relationships between performance in each component and overall performance in a race. The results provide three perspectives on the relationship between performance in each component of triathlon and the final placing in a race. These perspectives allow the identification of target split times that are required to achieve a certain final place in a race and the opportunity to make evidence-based decisions about race tactics in order to optimise performance.

Supporting athlete selection and strategic planning in track cycling omnium: A statistical and machine learning approach

This article describes the implementation of machine learning techniques that assist cycling experts in the crucial decision-making processes for athlete selection and strategic planning in the track cycling omnium. The omnium is a multi-event competition that was included in the Olympic Games for the first time in 2012. Presently, selectors and cycling coaches make decisions based on experience and opinion. They rarely have access to knowledge that helps predict athletic performances. The omnium presents a unique and complex decision-making challenge as it is not clear what type of athlete is best suited to the omnium (e.g., sprint or endurance specialist) and tactical decisions made by the coach and athlete during the event will have significant effects on the overall performance of the athlete. In the present work, a variety of machine learning techniques were used to analyze omnium competition data from the World Championships since 2007. The analysis indicates that sprint events have slightly more influence in determining the medalists, than endurance-based events. Using a probabilistic analysis, we created a model of performance prediction that provides an unprecedented level of supporting information that assists coaches with strategic and tactical decisions during the omnium.

The impact of semantic class identification and semantic role labeling on natural language answer extraction

In satisfying an information need by a Question Answering (QA) system, there are text understanding approaches which can enhance the performance of final answer extraction. Exploiting the FrameNet lexical resource in this process inspires analysis of the levels of semantic representation in the automated practice where the task of semantic class and role labeling takes place. In this paper, we analyze the impact of different levels of semantic parsing on answer extraction with respect to the individual sub-tasks of frame evocation and frame element assignment.

Mapping biomedical vocabularies: a semi-automated term matching approach.

Biomedical vocabularies vary in scope, and it is often necessary to utilize multiple vocabularies simultaneously in order to cover the full range of concepts relevant to a given biomedical application. However, as the number and size of these resources grow both redundancy (i.e., different vocabularies containing similar terms) and inconsistency (i.e., different terms in multiple vocabularies referring to the same entity) between the vocabularies increase. Therefore, there is a need for automatically aligning vocabularies. In this paper, we explore and propose new methods for detecting probable matches between two vocabularies. The methods build upon existing string similarity functions, enhancing these functions for the context of semi-automated vocabulary matching.