Bassam Hammo

Experimenting with a Question Answering System for the Arabic Language

The World Wide Web (WWW) today is so vast that it has become more and more difficult to find answers to questions using standard search engines. Current search engines can return ranked lists of documents, but they do not deliver direct answers to the user. The goal of Open Domain Question Answering (QA) systems is to take a natural language question, understand the meaning of the question, and present a short answer as a response based on a repository of information. In this paper we present QARAB, a QA system that combines techniques from Information Retrieval and Natural Language Processing. This combination enables domain independence. The system takes natural language questions expressed in the Arabic language and attempts to provide short answers in Arabic. To do so, it attempts to discover what the user wants by analyzing the question and a variety of candidate answers from a linguistic point of view.

Towards enhancing retrieval effectiveness of search engines for diacritisized Arabic documents

The majority of Arabic text available on the web is written without short vowels (diacritics). Diacritics are commonly used in religious scripts such as the holy Quran (the book of Islam), Al-Hadith (the teachings of Prophet Mohammad (PBUH)), children’s literature, and in some words where ambiguity of articulation might arise. Internet Arabic users might lose credible sources of Arabic text to be retrieved if they could not match the correct diacritical marks attached to the words in the collection. However, typing the diacritical marks is very annoying and time consuming. The other way around, is to ignore these marks and fall into the problem of ambiguity. Previous work suggested pre-processing of Arabic text to remove these diacritical marks before indexing. Consequently, there are noticeable discrepancies when searching the web for Arabic text using international search engines such as Google and yahoo. In this article, we propose a framework to enhance the retrieval effectiveness of search engines to search for diacritic and diacritic-less Arabic text through query expansion techniques. We used a rule-based stemmer and a semantic relational database compiled in an experimental thesaurus to do the expansion. We tested our approach on the scripts of the Quran. We found that query expansion for searching Arabic text is promising and it is likely that the efficiency can be further improved by advanced natural language processing tools.

Evaluation of Query-Based Arabic Text Summarization System

In this paper, we present and analyze the results of the application of Arabic query-based text summarization system - AQBTSS - in an attempt to produce a query-oriented summary for a single Arabic document. For this task, we adapted the traditional vector space model (VSM) and the cosine similarity measure to find the most relevant passages extracted form Arabic document to produce a text summary. We aim at using the short summaries in some natural language (NL) tasks such as generating answers for Arabic open domain question answering system (AQAS) as well as experimenting with categorizing Arabic scripts. The obtained results indicate that our simple approach for text summarization is promising.

Experimenting With Extracting Software Requirements Using NLP Approach

Requirement engineering is a fundamental step in the production of high quality software. Many attempts have been conducted to automate some aspects of the requirements engineering process. In this paper, we present a framework that provides the requirements engineers with an environment, which accepts English natural language requirements as input and automatically generates the corresponding UML class diagram designs. Moreover the framework can highlight the possibility of specification reusability through a reverse engineering process which saves the requirements engineers both time and efforts.

Dual Domain Watermarking in the Biological Color Model

In this paper we propose a novel method for watermarking. It uses the spatial and the discrete cosine transform (DCT) Domains. The proposed method deals with colored images in the biological color model, the Hue, Saturation, and Intensity (HSI). The method robustness is tested against several attacks. Watermark security is increased by using the Hadamard transform matrix. The watermarks used are meaningful, and of varying sizes.

WCBA: Weighted classification based on association rules algorithm for breast cancer disease

Abstract Breast cancer is the second most frequent human neoplasm that accounts for one quarter of all cancers in females. Among the other types of cancers, it is considered to be the main cause of death in women in most countries. An efficient classifier for accurately helping physicians to predict this chronic disease is in high demand. One approach for solving this problem has been tackled by many scholars using Association Classification (AC) techniques to enhance the classification process through applying association rules. However, most AC algorithms are suffering from the estimated measures used in the rule evaluation process and the prioritization techniques used at the attributes level, which could play a critical role in the rule generation process. In this article we attempt to solve this problem through an efficient weighted classification based on association rules algorithm, named WCBA. We also present a new pruning and prediction technique based on statistical measures to generate more accurate association rules to enhance the accuracy level of the AC classifiers. As a case study, we used WCBA to classify breast cancer instances with the help of subject matter experts from King Hussein Cancer Center (KHCC) located in Amman, Jordan. We compare WCBA with five well-known AC algorithms: CBA, CMAR, MCAR, FACA and ECBA running on two breast cancer datasets from UCI machine learning data repository. Experimental results show that WCBA, in most cases, outperformed the other AC algorithms for this case study. In addition, WCBA generates more accurate rules that contain the most efficient attributes for predicting breast cancer. WCBA algorithm aims to predict breast cancer in a patient. It serves all breast cancer patients by reducing the fear of the possibility of the recurrence of the disease and takes the necessary measures to prevent the progression of the disease and to predict breast cancer in a patient. The algorithm can be generalized to work on different domains with the help of subject matter experts.

Investigating the syntactic structure of Arabic sentences

In this paper, we employ the Government and Binding theory (GB) to present a system that analyzes the syntactic structure of some simple Arabic sentences structures. We consider different word orders in Arabic and show how they are derived. We include an analysis of Subject-Verb-Object (SVO), Verb-Object-Subject (VOS), Verb-Subject-Object (VSO), nominal sentences, nominal sentences with inna (or sisters), and question sentences. We use the analysis to develop syntactic rules for a fragment of Arabic, such that we include two sets of rules (1) rules on sentences structures that do not account for case and (2) rules on sentences structures that account for Noun Phrases (NPs) case. We present an implementation of the grammar rules in prolog. The results of testing the system t were reasonable with a high accuracy especially when the input sentences are tagged with identification of end cases.

WIT: Weka interface translator

Abstract Advancement in technology turns the big world into one small village. Regardless of what country you are living in, what language you are speaking or understanding, you should be able to benefit from the accumulated knowledge available on the Internet. Unfortunately, this is not the case with English being the de facto language of most programming languages, services, tools and web content. Many users are blocked from using these tools and services because they do not speak or understand English. Multilingual software evolved as a solution to this dilemma. In this paper, we describe the design and implementation of a user-friendly toolkit named Weka interface translator (WIT). It is dedicated to internationalize Weka, which is a collection of machine learning algorithms for data mining tasks widely used by many researchers around the world. WIT is a collaboration project between the Arabic natural language processing team from the University of Jordan and Weka’s development team from the University of Waikato. Its main goal is to facilitate the translation process of Weka’s interfaces into multi-languages. WIT is downloadable through SourceForge.net and is officially listed on Weka’s wiki spaces among its related projects. To experiment with WIT, we present Arabic as a pilot test among many languages that could benefit from this project.

Experimenting with Arabic text visualizing

The amount of unstructured textual data on the Internet has been increased dramatically. Text visualization becomes a significant tool that facilitates knowledge discovery and insightful presentation of large amounts of data. In this paper we present a technique of the visual exploration of Arabic text documents. We apply Latent Semantic Indexing (LSI) as a dimensionality reduction technique that helps in extracting some knowledge from text documents. We present an Arabic Visualization System. The experiments were carried on several datasets extracted from an Arabic corpus. The system can create views from different perspectives based on the users needs. It is successfully used for visualizing different kinds of document corpora. The system is very helpful for data analysis offering quick insight into the structure of the visualized corpus.

Identifying β-thalassemia carriers using a data mining approach: The case of the Gaza Strip, Palestine

Thalassemia is considered one of the most common genetic blood disorders that has received excessive attention in the medical research fields worldwide. Under this context, one of the greatest challenges for healthcare professionals is to correctly differentiate normal individuals from asymptomatic thalassemia carriers. Usually, thalassemia diagnosis is based on certain measurable characteristic changes to blood cell counts and related indices. These characteristic changes can be derived easily when performing a complete blood count test (CBC) using a special fully automated blood analyzer or counter. However, the reliability of the CBC test alone is questionable with possible candidate characteristics that could be seen in other disorders, leading to misdiagnosis of thalassemia. Therefore, other costly and time-consuming tests should be performed that may cause serious consequences due to the delay in the correct diagnosis. To help overcoming these challenging diagnostic issues, this work presents a new novel dataset collected from Palestine Avenir Foundation for persons tested for thalassemia. We aim to compile a gold standard dataset for thalassemia and make it available for researchers in this field. Moreover, we use this dataset to predict the specific type of thalassemia known as beta thalassemia (β-thalassemia) based on hybrid data mining model. The proposed model consists of two main steps. First, to overcome the problem of the highly imbalanced class distribution in the dataset, a balancing technique called SMOTE is proposed and applied to handle this problem. In the second step, four classification models, namely k-nearest neighbors (k-NN), naïve Bayesian (NB), decision tree (DT) and the multilayer perceptron (MLP) neural network are used to differentiate between normal persons and those patients carrying β-thalassemia. Different evaluation metrics are used to assess the performance of the proposed model. The experimental results show that the SMOTE oversampling method can effectively improve the identification ratio of β-thalassemia carriers in a highly imbalanced class distribution. The results reveal also that the NB classifier achieved the best performance in differentiating between normal and β-thalassemia carriers at oversampling SMOTE ratio of 400%. This combination shows a specificity of 99.47% and a sensitivity of 98.81%.