Sharif Digital Repository

Automated Planning is an active research area in Artificial Intelligence. In Classical planning, for simplicity, time is considered as the order of actions in plan. In temporal planning, due to the importance of time in real world problems, this simplifying assumption is not considered, and time is explicitly used in the planning process. Most of current methods for temporal planning are extensions of classical planning methods to include the explicit definition of time. Planning using Satisfiability is used as an efficient method to find optimal solutions for classical planning problems. In this dissertation, a temporal planner based on Satisfiability has been developed. This planner, as we... 

Automated planning is a branch of artificial intelligence that studies intelligent agents’ decision making process.In planning, we can design agents that can decide on their own, about how to perform tasksthat are assigned to them. In classical planning, there is a restrictive assumption that actions in plans are totally ordered. By relaxing this restrictive assumption, partial order planninghas been created.Partial order planning uses a general principle, called the least commitment principle that results in a better performance than other classical planning methods. Yet, this branch of planning cannot compete with newer planning methods like heuristic search planning.That is why there has... 

Automatic grammar induction is one of attractive research topics in natural language processing field. Automatic grammar induction methods can be categorized into three main groups of supervised, semi-supervised and unsupervised methods based on the type of training data that they need. Unsupervised methods are more difficult than two other. Data Oriented Parsing (DOP) is one of successful methods in unsupervised group. This method has been trained by some examples of language as same as child, then it parses new sentences based on its training knowledge. The aim of this project is finding and improving performance of UDOP method on Persian language as a Free Word Order language. Results of... 

Temporal relation classification is one of the contemporary demanding tasks in natural language processing. This task can be used in various applications such as question answering, summarization, and language specific information retrieval. Temporal relation classification methods can be categorized into three main groups of supervised, semi-supervised, and unsupervised (based on the type of the training data that they need). In this thesis, we have two main goals: first, improving accuracy of temporal relation learning, and second, decreasing supervision of algorithm as much as possible. For achieving these goals, three main steps are proposed. In the first step, we propose an improved... 

The news headlines should represent the main and the most important topics of their stories. The task of selecting an appropriate headline for news stories is mainly done by journalists. The goal of this project has been the design and implementation of a system to automate this task, that is generating headlines for news. This task has been done for Persian news stories. There are various methods for automatic headline generation in English and some other languages, but no work has been done for Persian, yet. Thus, we have adopted some of the ideas from those methods, and do the remaining by our initiation. Our proposed method consists of three main parts: keyword extraction, most important... 

As digital content grows rapidly due to the internet, user reviews about different topics such as product quality can be used as a rich source to check and analyze product quality and performance. Automatic methods are being widely used to extract these information because of the massive amount of available resources. Sentiment analysis is one of the important fields in natural language processing, which uses a combination of learning and rule-based methods to extract subjective information out of documents. Aspect based sentiment analysis deals with sentiment analysis based on each aspect of the product. It consists of two main steps: first, aspects should be extracted from the reviews and... 

The Question-Answering(QA) problem has long been a significant focus of researchers. Its connection with natural language understanding and knowledge retrieval makes it one of the most critical issues in Natural Language Processing (NLP). Given the inefficiency of simple question-answering methods, multi-hop question-answering (Multi-hop QA) across multiple documents has become one of the most attractive problems in recent years. In general, multi-hop question-answering is supposed to answer natural language questions that require extracting and combining information conained in several documents and performing reasoning about that information. The ability to answer questions and perform... 

Planning is an important branch of Artificial Intelligence. It is the area of study concerned with the automatic generation of plans to solve problems within a particular domain. At its simplest case, a plan is a total ordered sequence of actions. Given an initial state, planner tries to find the actions required to achieve some desired goal conditions. Planning technology has been successfully used in a variety of applications, including NASA's Remote-Agent Planner/Scheduler, which was practically employed in the Deep Space 1 spacecraft, satellite planning and scheduling for the European Meteostat, planning in a forest fire simulation system, etc. AI planning encompasses some difficulties... 

Event extraction is one of the important tasks in Natural Language Processing (NLP). Many NLP applications like question answering, information extraction and summarization need to have some knowledge about events of input documents. There are several definitions for events in NLP domains. In this dissertation, the event is viewed as an element in a network of temporal information. Therefore, the project has been based on the ISO-TimeML specification language, which is the standard scheme for temporal information processing in natural texts. Event extraction based on ISO-TimeML has been performed for a number of languages including English, French, Spanish, and Korean. However, for Persian... 

Automated planning is a branch of Artificial intelligence aimed at obtaining plans (i.e. sequences of actions) for solving complex problems or for governing the behavior of intelligent agents. Reduction of planning problems to satisfiability problems is one of the most successful approaches to automated planning. In this method first the planning problem with a preset length is coded to a satisfiability problem. Then, a general SAT solver is used for solving the obtained SAT formula. To find the plan with optimum length, plans with increasing coding lengths are evaluated by a SAT solver in a sequential manner. Since much of descriptive information of planning problems is lost in converting... 

Planning is one of the most important branches of artificial intelligence. The goal of planning is automated production of sequence of actions (plans) for some specified tasks. However the computation done by planners is very time consuming and complicated. Distributing computation is one solution to overcome such difficulties. Distribution divides the load of process between a number of agents and increases robustness. The purpose of this project is to reduce the computational overhead of planning by the means of distributing the planning process between a numbers of planners. In the proposed method, we first divide a classical planning problem to two separate subproblems by decomposition... 

Named entity recognition (NER) is one of important tasks in natural language processing (NLP). Named entities consist of specific nouns such as personal names, organizations, locations, etc., which refer to important entities in text. NER contributes towards other NLP tasks such as machine translation, text summarization ,and text classification. In the recent decade, with respect to development of deep learning (DL) methods, considerable progress has been made in this field. The objective here is to propose an efficient method for NER in Farsi (Persian) text through DL methods. Since deep neural networks require a great deal of training data, and due to the fact that Farsi lacks such data,... 

Temporal planning is one of the branches of AI planning, which has attracted a considerable amount of attention in recent years due to the importance of temporal properties in real world planning problems. Like in many other AI fields, there exists an important trade-off in temporal planning between the variety of solvable problems and the efficiency of the planner. Most of the previously proposed temporal planners use heuristic state-space search method. This method, despite being quite efficient, is not complete for so-called “problems with required concurrency”. On the other hand, plan-space based and satisfiability based temporal planners, which can tackle problems with required... 

Grammar induction is an important area of natural language processing. There are two general methods for recognizing the syntactic structure: constituency and dependency parsing. The unique nature of the dependency parsing, in which the word order does not affect the syntactic structure of the sentence, make it an appropriate option for parsing free-word-order languages such as the Persian. In this thesis, dependency-based methods are used to parse Persian sentences. Manual induction of the grammar is a time-consuming and tedious task. However, machine learning algorithms facilitated this task to a great deal. One of the most effective algorithms in this field is the deep neural networks... 

Planning, as one of the major area of the artificial intelligence, has substantial applications in different science and technology fields. In the classical planning, usually some restricted assumptions are made to simplify the planning procedure. For example, for the world representation, states and actions are defined precisely; however, in the real world such information is mostly inacurate. Therefore, in order to address this inacuracy, flexible planning has been developed. In this technique, robust restrictions such as complete satisfaction of preconditions are interpreted as preferences or priority. In other words, the flexible planning aims at fulfilling planning goals with maximum... 

Question Answering (QA) systems have reached human-level accuracy; however, these systems are vulnerable to adversarial examples. Recently, adversarial attacks have been widely investigated in text classification. However, there have been few research efforts on this topic in QA systems. In this thesis our approach is improving the robustness of QA systems using deep neural networks. In this thesis, as the first proposed approach, the knowledge distillation method is introduced to create a student model to improve the robustness of QA systems. In this regard, the pre-trained BERT model was used as a teacher, and its impact on the robustness of the student models on the Adversarial SQuAD...