Abstract Dictionaries are rich sources of lexical information about words that is required for many applications of human language technology. However, publishers prepare printed dictionaries for human usage not for machine processing. This paper presents a method to structure partly a machine-readable version of the Arabic-English Al-Mawrid dictionary. The method converted the entries of Al-Mawrid from a stream of words and... Dictionaries are rich sources of lexical information about words that is required for many applications of human language technology. However, publishers prepare printed dictionaries for human usage not for machine processing. This paper presents a method to structure partly a machine-readable version of the Arabic-English Al-Mawrid dictionary. The method converted the entries of Al-Mawrid from a stream of words and punctuation marks into hierarchical structures. The hierarchical structure or tree structure expresses the components of each dictionary entry in explicit format. The components of an entry are subentries and each subentry consists of defining phrases, domain labels, cross-references, and translation equivalences. We designed the proposed method as cascaded steps where parsing is the main step. We implemented the parser through the parsing expression grammar formalism. In conclusion, although Arabic dictionaries do not have microstructure standardization, this study demonstrated that it is possible to structure them automatically or semiautomatically with plausible accuracy after inducing their microstructure. Read More Read Less

Abstract A Hadith is a report of the deeds or sayings of the prophet Muhammad. Each of these reports were orally transmitted from one person to another till it reached a person who recorded the report along with the chain of transmission. We present a system to automatically extract the chain of narrators from a Hadith through Named Entity Recognition and... A Hadith is a report of the deeds or sayings of the prophet Muhammad. Each of these reports were orally transmitted from one person to another till it reached a person who recorded the report along with the chain of transmission. We present a system to automatically extract the chain of narrators from a Hadith through Named Entity Recognition and Classification, and present these transmission chains as a network. In a Hadith, the name of a person may appear as a narrator or as someone who is mentioned in the Hadith. This distinction of names is important as identifying and evaluating the narrators is an important part of Hadith studies. We manually annotated a large Hadith corpus with named entities and used a set of keywords and special verbs to train machine learning algorithms for named entity recognition and classification. The keywords and special verbs identified the context surrounding the tokens labeled as named entities. We compared the performance of different classifiers including generative (Naïve Bayes), and discriminative (K-nearest neighbour and decision tree) and were able to achieve a 90% precision and 82% recall for the named entities. The classifiers were evaluated on a different corpus within the same domain that resulted in an 80% precision and 73% recall. The best classifier was used to label a bigger Hadith corpus and the narrators names thus identified from each Hadith were concatenated to create a chain of narration from the Hadith. These chains were represented as a graph of narrators in the end. Read More Read Less

Abstract In the study of content authentication and tamper detection of digital text documents, there are very limited techniques available for content authentication of text documents using digital watermarking techniques. In this paper, we have extended the proposed LNMZW3 algorithm presented in [27] for content authentication and tamper detection of English text documents and abbreviated as ADV-LNMZW3. In the enhanced algorithm... In the study of content authentication and tamper detection of digital text documents, there are very limited techniques available for content authentication of text documents using digital watermarking techniques. In this paper, we have extended the proposed LNMZW3 algorithm presented in [27] for content authentication and tamper detection of English text documents and abbreviated as ADV-LNMZW3. In the enhanced algorithm , third order and letterbased of Markov model has uses in this paper as soft computing tools in order to English text analyse in order to find the inter-relationships and utilize theses features to generate and detect a watermark in order to identify the status of text document such as authentic, or tampered. One of the important enhancements has done on LNMZW3 algorithm and presented in this paper is study the performance average of our enhanced algorithm named as ADV-LNMZW3 against different sizes of datasets and compare them with two modern proposed algorithms named LNMZW1 and LNMZW2. The study of dataset size effect also has been examined in this paper against common volumes of insertion, deletion and reorder attacks. The enhanced algorithm (ADV-LNMZW3) was implemented using PHP Programming language with Net Beans IDE 7.0. Furthermore, the effectiveness and feasibility of our ADV-LNMZW3 algorithm has proved and compared with other recent algorithms with experiments using five datasets of varying lengths and different volumes of attacks. The experiment showed that the ADV-LNMZW3 algorithm had better performance and robustness, and it is more secure than other algorithms especially in case of insertion and deletion attacks.The effect of document size on watermark robustness was examined. The results showed that the watermark robustness is enhanced with small and medium documents size. The results showed that our ADV-LNMZW3 algorithm was applicable for all sizes of text document, and it is recommended for tampering detection against small, and medium sizes of text documents. However, the LNMZW1 was found to be applicable under large size of text documents. Read More Read Less