Sharif Digital Repository

The aim of this paper is to analyze the Farsi-English statistical machine translation systems as a useful communication tool. Improvement of the nation's communication increases the need of easier way of translating between different languages in front of expensive human translators. In this work, a statistical phrase-based system is run on Farsi - English pair languages and the effect of its parameters on the translation quality has been deeply studied. Using BLEU as a metric of translation accuracy, the system achieves an improvement of 1.84%, relative to the baseline accuracy, which is increment from 16.97% to 18.81% in the best case  

One of the most primary phases of automatic text processing is spelling error detection and grapheme normalization. Storing textual documents faces several problems without passing this phase, which causes a disturbance in retrieving the documents automatically. Therefore, specialists in the fields of natural language processing and computational linguistics usually make an attempt to sample various data through presenting ideal methods and algorithms in order to reach the normalized data. Several researches have been conducted on English and some other languages, which have been followed by a certain amount of researches on Farsi too. Sometimes, these several researches have remained to be... 

In this paper we try to alleviate the problem of scares resources for developing Farsi-English Statistical Machine Translation system (SMT). It is done by applying Active Learning (AL) idea to choose more informative sentences to be translated by a human and then be added to the base-line corpus. While using the human translations is worthless in compare to the other approaches of corpus gathering (like automatic approaches), it is more costly too. So, in this way we can improve the translation system with less cost. This is done in intricate to human translator. Applying Active learning idea to a SMT system, changes it to a system which can improve its based-line corpus by asking for the... 

Despite many advances in optical character recognition in general, there are still serious challenges remaining in recognizing Farsi text. The main reason is the cursive nature of the letters in written Farsi, i.e., depending on the position of a letter within a word, it might join to its neighboring letters, which consequently changes the shape of the character. As a result, each letter can have up to four different character shapes. In addition to the problem of segmenting the characters, the increased number of characters makes the recognition task even more challenging. This paper introduces a complete framework for character recognition, including a method for segmenting the characters... 

With increasing use of computers in everyday life, improved communication between machines and human is needed. To make a right communication and understand a humankind face which is made in a graphical environment, implementing the audio and visual projects like lip reading, audio and visual speech recognition and lip making are needed. Lack of a complete audio and visual database for this application in Farsi language made us provide a new complete Farsi database for this project that is called SFAVD. It is a unique audio and visual database which in addition to considering Farsi conceptual and speech structure, it considers influence of speech on lip changes. This database is created for... 

Currently there is no standard test collection for evaluation of Farsi information retrieval systems. In this paper we introduce Mahak, the first complete test collection generally available for evaluating Farsi information retrieval systems. In addition, we have used different methods for constructing Mahak qrels and we have compared performance of these methods  

With the growing presence of computers in everyday life, communication improvement between human and machines is inevitable. Talking faces are the faces whose movements are synchronized to speech. They have an effective role in many applications. Lip is the most important part of a talking face. The main goal of this project is implementing a natural and human-like lip movement synthesis system for the Farsi language. For this purpose, a comprehensive audio visual database called SFAVD1 was designed and used. After extracting the sufficient features and designing a parallel Hidden Markov Model, the speech driven lip movement sequence generator system for Farsi input speech was implemented.... 

In this paper an off-line ArabiclFarsi handwritten recognition Algorithm on a subset of Farsi name is proposed. In this system, There is no sub-word segmentation phase. Script database includes 3300 images of 30 Farsi common names. The features are wavelet coefficients extracted from smoothed word image profiles in four standard directions. The Centers of competitive layer of RBF neural network have been determined by combining GA and K-Means clustering algorithm. Weights of supervised layer has been trained by using LMS rule and the distances of feature vector of each sample to the centre of RBF network have been computed based on warping function. Experimental results show advantages of...