M.Phil / MS

Now showing 1 - 2 of 2

Multi-label Financial Text Classification: Corpus and Methods
(Library Information Services, COMSATS University Islamabad, Lahore Campus, 2023) Laraib Khalid; FA21-RCS-018; LHR TP 8460; Dr. Jawad Shafi
Multi-label financial text classification is evolving but widely researched Natural Language Processing (NLP) task of the current decade. In the recent past, several research efforts have been reported for financial text classification for English and other European languages for instance. Moreover, they are performed for single-label classification (i.e., a document/text is labeled with a single label). However, there is a dearth of South Asian languages, particularly Urdu. Which has around 300 million (non)native speakers around the world and whose digital text is increasing day by day. Unfortunately, very less research efforts have been addressed for Urdu NLP, for instance, financial text classification. Therefore, this research work addresses this research gap for the Urdu language by developing standard evaluation resource and applying baseline supervised and deep learning techniques using multi-label (document/text is labeled with multi labels) classifiers on the proposed corpus. Our proposed Urdu Multi-label Financial Text Classification (UML-FTC-23) corpus contains 24,340 documents and contains different genres and text of business domain. The UML-FTC-23 has been manually annotated with 13 labels ensuring detailed and comprehensive classification of the financial documents/text. It is worth noting that each document in our corpus has been assigned a minimum of two labels and a maximum of nine labels to understand the fine-grained analysis of the financial text and to provide the multi-label classification of text. To demonstrate the quality of the proposed corpus, UML-FTC-23 has been evaluated for multi-label financial text classification by using baseline approaches: 1- Bag of Words (BoW) n-gram, and 2- Term Frequency-Inverse Document Frequency (TF-IDF). Additionally, we have used eight different multi-label classifiers on extracted features. The proposed techniques have been evaluated using several multi-label evaluation measures for instance, exact match, hamming loss, and F1. Furthermore, we have used deep learning methods, including CNN, LSTM, GRU, and RNN on UML-FTC-23 corpus. Result demonstrates that the supervised TF-IDF technique with the multi-label binary relevance classifier achieves the best F1 score of 0.89. Moreover, on the best result using deep learning method are with the CNN which produces the 0.81 F1 score. This indicates that the proposed corpus and techniques are of worth and can be used in various research fields. The proposed corpus and resources are freely available for academic research purposes.
Detecting Real, Simulated and Artificial Cases of Paraphrasing
(Library Information Services COMSATS University Islamaabad Lahore Campus, 2021) Ayesha Shahzadi; SP19-RCS-016; Dr. Jawad Shafi,
Paraphrase detection is the process of identifying the use of existing text(s) as a new text in the same context with different alterations/modifications. In recent years, paraphrase detection has gained the attention of the research community due to its potential applications in different domains of NLP and Machine Learning. Paraphrasing may carry out on three use case levels 1) Real Cases, 2) Simulated Cases, and 3) Artificial Cases. Real Cases are examples from the real-world data available on different platforms, websites, e books, and over the internet on multiple topics and domains. When human manipulates the data content by using different grammatical rules and editions for paraphrasing and then claims to be original data that is Simulated Cases paraphrasing. Now the world is shifting from manual tasking to the availability of tools to modify data freely on the internet. Modifications in available data by using online tools is an example of Artificial Cases paraphrasing. To identify the use case of paraphrasing is a crucial task because, without any computational and experimental environment, no tool can identify the paraphrasing type. More importantly, Artificial Cases paraphrase detection is the most difficult task to identify. The world is lacking the standard set of examples for Simulated Cases and Artificial Cases. To fulfill this gap, we purposed one Simulated Cases corpus and three Artificial Cases corpora for paraphrasing in the English language on Sentence Level with binary classification as paraphrased or non-paraphrased. To address this problem, this study has three main goals: (1) develop benchmark corpora for simulated and Artificial Cases of paraphrasing and (2) apply WordNet-based approaches, Kull-back Libeler, and approach on Real, Simulated, and Artificial Cases of paraphrasing and (3) identify which automatic paraphrasing tools are more difficult to detect paraphrasing. We tried to contribute on the initial level, we developed a corpus for Simulated Cases and Artificial Cases from previous studies. We developed benchmark corpora as Quora Simulated Corpus on Simulated Cases, Artificial Article Rewriter Corpus, Artificial ix Rewriter Tool Corpus, and Artificial Paraphrasing Tool Corpus on Artificial Cases. Each corpus contains 5801 sentence pairs for the English language from different domains. We applied different approaches like N-gram Overlap. Kull-back Liebler and Wordnet based approaches to extract features from data on Real, Simulated, and Artificial Cases corpora. Five different machine learning classifiers like Random Forest, Decision Tree, Multilayer Perceptron, Adaboost, and Gradient Boosting classifiers were evaluated by using Precision, Recall, and F1 measure. We only reported the highest F1 scores, as AARC, ARTC, and QSC scored 85%, APTC scored 89%, and MSRPPC scored 71% results for all combined approaches.a

M.Phil / MS

Browse

Filters

Settings

Sort By

Results per page

Search Results