Search or add a thesis

Advanced Search (Beta)
Home > Generic Urdu Nlp Framework for Urdu Text Analysis: Hybridization of Heuristics and Machine Learning Techniques

Generic Urdu Nlp Framework for Urdu Text Analysis: Hybridization of Heuristics and Machine Learning Techniques

Thesis Info

Access Option

External Link

Author

Khan, Wahab

Supervisor

Ali Daud

Program

PhD

Institute

International Islamic University

City

Islamabad

Province

Islamabad.

Country

Pakistan

Thesis Completing Year

2019

Thesis Completion Status

Completed

Subject

Computer Science

Language

English

Link

http://prr.hec.gov.pk/jspui/bitstream/123456789/10445/1/Wahab%20Khan_CS_2019_IIU_Incomp.pdf

Added

2021-02-17 19:49:13

Modified

2024-03-24 20:25:49

ARI ID

1676727764317

Asian Research Index Whatsapp Chanel
Asian Research Index Whatsapp Chanel

Join our Whatsapp Channel to get regular updates.

Similar


The internet was initially designed to present information to users in English. However, with the passage of time and the development of standard web technologies such as browsers, programming languages, libraries, frameworks, databases, front and back-ends, protocols, APIs, and data formats, the internet became a multilingual source of information. In the last few years, the natural language processing (NLP) research community has observed a rapid growth in online multilingual contents. Thus, the NLP community maims to explore monolingual and cross-lingual information retrieval (IR) tasks. Digital online content in Urdu is also currently increasing at a rapid pace. Urdu, the national language of Pakistan and the most widely spoken and understandable language of Indian sub-continent, is considered a low-resources language (Mukund, Srihari, & Peterson, 2010). Part of speech (POS) tagging and named entity recognition (NER) are considered the most basic NLP tasks. Investigation of these two tasks in Urdu is very hard. POS tagging, the assignment of syntactic categories for words in running text is significant to natural language processing as a preliminary task in applications such as speech processing, information extraction, and others. Named entity recognition (NER) corresponds to the identification and classification of all proper nouns in texts, and predefined categories, such as persons, locations, organizations, expressions of times, quantities and monetary values, etc. it is considered as a sub-task and/or sub-problem in information extraction (IE) and machine translation. NER is one of the hardest task in Urdu language processing. Previously majority Urdu NER systems are based on machine learning (ML) models. However, the ML model needs sufficiently large annotated corpora for better performance(Das, Ganguly, & Garain, 2017). Urdu is termed as a scared resource language in which sufficiently large annotated corpus for ML models’ evaluation is not available. Therefore, the adoption of semi-supervised approach which is largely dependent on usage of the huge amount of unlabeled data is a feasible solution. In this thesis, we propose a generic Urdu NLP framework for Urdu text analysis based on machine learning (ML) and deep learning approaches. Initially, we addressed POS challenges by developing a novel tagging approach using the linear-chain conditional random fields (CRF). We employed a strong, stable, balanced language-independent and language dependent feature set for Urdu POS task and used the method of context words window. Our approach was evaluated against a support vector machine (SVM) technique for Urdu POS - considered Abstract WAHAB KHAN Reg: No. 72-FBAS/PHDCS/S12 vi as the state of the art - on two benchmark datasets. The results show our CRF approach to improving upon the F-measure of prior attempts by 8.3 to 8.5%. Secondly, we adopted deep recurrent neural network (DRNN) learning algorithms with various model structures and word embedding as a feature for the task of Urdu named entity recognition and classification. These DRNN models include long short-term memory (LSTM) forward recurrent neural network (RNN), LSTM bi-directional RNN, backpropagation through time (BPTT) forward RNN and BPTT bi-directional RNN. We consider language-dependent features such as part of speech (POS) tags as well as language independent features such as N-grams. Our results show that the proposed DRNN-based approach outperforms existing work that employ CRF based approaches. Our work is the first to use DRNN architecture and word embedding as a feature for Urdu NER task and improves upon prior attempts by 9.5% in the case of maximum margin.
Loading...
Loading...

Similar Books

Loading...

Similar Chapters

Loading...

Similar News

Loading...

Similar Articles

Loading...

Similar Article Headings

Loading...

صباح الدین عمر

صباح الدین عمر
افسوس ہے کہ اردو کے ایک عاشق و شیدائی جناب صباح الدین عمر کا انتقال ہوگیا، وہ لکھنؤ کی روایات کے بڑے دلدادہ اور اس کی تہذیب و ثقافت کا نمونہ تھے، وہ سرکاری ملازم تھے، یوپی کے محکمہ اطلاعات کے اردو ماہنامہ ’’نیادور‘‘ کے ایڈیٹر بھی رہے، اترپردیش اردو اکادمی کے قیام کے بعد اس کے سکریٹری ہوئے اور اس کا رسالہ اکادمی ان کی ادارت میں شائع ہوا، ریٹائرڈ ہونے کے بعد اردو اکادمی اور فخرالدین علی احمد میموریل کمیٹی کے برابر رکن رہے اور ان کو اپنے مشوروں اور تجربوں سے بڑا فائدہ پہنچایا، طبعاً شریف اور مخلص تھے، دوسروں کی مدد کرکے خوشی محسوس کرتے تھے، اﷲ تعالیٰ اردو کے اس عاشق و خادم کی مغفرت فرمائے، آمین!! (ضیاء الدین اصلاحی۔ دسمبر ۱۹۹۱ء)

 

الحديث الضعيف وما يتعلق به من الأحكام

Legitimation among scholars, since they fall to category of hadith dho’if (weak). Therefrom, several scholars argued that we might use them for hujjah mutlaq (absolute argumentation), while some others said it might be wiser not to use them at all. Yet there is also another opinion which said it could be used under special conditions. Based on this, this study aims to uncover and shed light the disagreements above scientifically, as well as to find he differences and the influence of the jurisprudence of law-making (fiqh). Then, the researchers sought to raise a strong opinion based on the arguments presented in the thesis, so which the researchers and or anyone who wants to practice the Hadith may find helpful.

School Principal As a Strategic Leader: A Case-Study of a Private Sector School in District Chitral in Khyber Pakhtunkhawa Kpk Pakistan

School leadership plays a significant role in determining the levels and quality of school improvement process and its outcomes. The position of the school principal is of a door keeper to change that means how he receives the guests and what are the arrangements to welcome and how to please the guests. The school culture, practices and structure mainly revolve around the principals’ qualities, specifically in the private educational systems. The educational reforms and the fast moving technological era demand that the school leaders can only survive if they have strategic leadership qualities. They also require a strong vision for the institution. To achieve this vision, there should be a strong mission with strategic working styles and measures. The leaders should have the vision, the ability to share the vision with people inside as well as outside the organisation and inspire them to work jointly to accomplish the vision. Strategic leaders are those who have different thinking powers than the ordinary leaders and far-sighted lenses and thought processes to any endeavor. They scan the current position of the institution with respect to its internal and external environment, anticipate change proactively, set goals and objectives, align the organisation’s culture to these goals and negotiate across a wide breath of stakeholders for the survival of the organisation and its continuous improvement. They reflect on the strategies used for improvement and sustain the change in the complex and uncertain changing context. The purpose of the study is to explore and understand the principal’s role as a strategic leader. It also aims at understanding the role of the principal’s practices and position with futuristic lens. Moreover, to understand the principal’s powers to compete in a competitive situation in rural private context and the strategies used to sustain in the education market. The study employed the exploratory case study method selecting one school in the rural context of Chitral, Khyber Pakhtunkhawa (KPK), Pakistan. Qualitative data gathering strategies, namely; document analysis, observations, semi-structured and focus group discussions were made use of. The data were categorised and classified on the basis of the themes focusing on the principal’s practices and functions. The study findings reveal that the principal’s role is essential for strategic development in schools and strategic planning is fundamental to move towards the strategic direction. Strategic role of the principals is important for the educational development in far-flung rural areas. This study may help educational