Sharif Digital Repository

Text classification is one if the widest and most important applications in data mining. Because of the huge number of features in these applications, a method for dimensionality reduction is needed before applying the classification algorithm. Various number of methods for dimensionality reduction and feature selection are proposed. Feature selection based on feature ranking has received much attention by researchers. The major reasons are their scalability, ease of use, and fast computation. Feature ranking methods are divided to different categories and use different measures for scoring features. Recently ensemble methods have entered the field of ranking, and achieved more accuracy... 

Author identification using statistical methods is a branch of authorship attribution which is one of important problems in natural language processing. Using different statistical methods, an anonymous text is attributed to an author. One of the primary parts of the task is to choose the appropriate stylistic features of the text in order to study the significances of style. These features must be quantitatively studied and could be extracted in lexical level, character level, and syntactic or semantic levels. The next step is text classification in which different machine learning methods such as decision tree, Artificial Neural Networks, Naïve Bayes and other methods could be used.... 

Recognizing human actions is an important and challenging topic in computer vision, which has important applications such as video surveillance and Indexing. From a computational perspective, actions can be defined as three-dimensional patterns, in space and in time which can be modeled using several representations. Action representations differ in visual information used in spatial dimensions (e.g., shape or appearance) and the representation of dynamics in time. The goal of this thesis is to develop new techniques and improve current results in action categorization. As such, using a general structure, three methods are proposed. In this structure, local spatio-temporal features are... 

According to modal interpretations of quantum mechanics, a state possesses definite values even if it is not an eigenstate. On the other hand, Kochen-Specker theorem holds that all possible possessed values by a system can not be definite. Consequently, modal interpretations to block this theorem, take new metaphysics of properties. In this dissertation, I will show that dispositionalism is that metaphysics. According to dispositionalism, causal powers of a property are constituents of the property. As regards this outline, the thesis would be divided into 5 chapters. In chapter one, presuppositions of later arguments are clarified. Chapter two is related to algebra of properties and finally... 

One of the new paradigms in information retrieval is to develop textual Question-Answering systems. Question-Answering (QA) is an advanced IR process at which for a natural language question, the answer is extracted and issued in natural language. The QA systems are divided into two general groups: Open-Domain QA and Restricted-Domain QA.
In this research field, a number of different models and methods are developed in which a document collection is used to retrieve candidate answers and then different methods are deployed to detect and eliminate irrelevant ones from answer set. Most of these methods decide based on expected semantic answer type, which is determined using pre-defined... 

Studies show that a large proportion of sent emails are spam. Spam is one of the major problems of e-mail users that result in wasting time and cost. To overcome this problem different ways are used, one of the best ways is detecting spam based on their contents. Separating legitimate e-mails and spam within their contents can be categorized as text classification. So machine-learning approaches are extremely applied in text classification, that machine-learning algorithms can be used for spam classification. However, in the majority of these algorithms, training phase is in a batch. Whereas using incremental learning algorithms is preferred in many applications, especially spam detections.... 

In recent years, ability of computers to recognize human actions, because of numerousapplications, has attracted scientists. Surveillancesystems in house, work and public places, human computer interaction, study of human movement problems, remote supervision of ill or old people and sport training are only some of the applications. In this thesis 10 actions are considered. These actions are Walking, Running, Galloping side, Bending, Jump jacking, Jumping, Jumping in place, Skipping, Waving one hand and Waving two hands. All actions exist in Weisemann dataset so this dataset is used as training and testing dataset. Here important objectives are recognising human action so that it is... 

Due to immense increase in availability of text data, interest in using machine learning models to solve problems previously impossibly costly has increased significantly. The first step is to represent natural language in a form that is easy for the machine learning algorithms to work on. Recent advances in learned representation of text data using simple neural networks(e.g. word2vec and doc2vec) helped increase performance of natural language processing on downstream tasks. Here we show that methods like doc2vec that were examined mostly in the English language can be used on Persian(Farsi) with little modification. To Demonstrate this, we use text classification tasks, and train... 

Nowadays spam has become as a universal problem which all email users are familiar with it. Studies show that a large proportion of sent emails are spam. Obviously it results in wasting a vast range of resources. There is different ways to fight spam; each of them has its own strengths and weaknesses. The most common filtering technique is content based filtering. This problem has been addressed as a text classification problem. Two main defect of spam filtering techniques are manually definition of rules and circumventing them, one solution for overcoming this problem is applying machine learning algorithms. Spam classification using machine learning techniques is very successful and... 

Simultaneous development of technology and social networks and universal access to them caused to produce and distribute huge volume of videos that recognition of their content without use of machine vision is very hard. This thesis examine some video classification algorithms to improve them. The algorithm that is used to improve is based on one of local descriptor algorithms. At first with using STIP tools, the local interest point found by Harris3d and describe by HOG/HOF. Then by using Bag of Features, all local descriptors in a video produce a descriptor per videos. Bag of Features divide the domain of all local descriptors from all videos to K cluster and produce a vector per video...