The automatic generation of sentences is a domain of Natural Language Processing (NLP); it is placed in the middle of computer science and linguistics. This is a very complex discipline; the aim of this is to create automatically correct sentences from a list of words which can serve as the basis for such various applications such as automatic translation, question-answering systems, correcting syntactic errors and so on. In this study, we present the use of the linguistic approach of Chomsky's minimalist grammar. We begin with the elaboration of the lexicon that constitutes the essential link of the generation. Then, based on this lexicon, we treat the merge and move operations to build a syntactically correct sentence.

Stylometry plays an important role in the intrinsic plagiarism detection, where the goal is to identify potential plagiarism by analyzing a document involving undeclared changes in writing style. The purpose of this paper is to study the interaction between syntactic structures, attention mechanism, and contextualized word embeddings, as well as their effectiveness on plagiarism detection. Accordingly, we propose a new style embedding that combines syntactic trees and the pre-trained Multi-Task Deep Neural Network (MT-DNN). Additionally, we use attention mechanisms to sum the embeddings, thereby experimenting with both a Bidirectional Long Short-Term Memory (BiLSTM) and a Convolutional Neural Network (CNN) maxpooling for sentences encoding. Our model is evaluated on two sub-task; style change detection and style breach detection, and compared with two baseline detectors based on classic stylometric features.

Syntax plays a key role in natural language processing, but it does not always occupy an important position in applications. The main objective of this article is to solve the problem of the grammatical case ending errors produced by Arabic learners or certain common errors. Arabic can be considered more complex than English or French. He does not have vowels; diacritic signs (vowels) are placed above or below the letters. These diacritic signs are abandoned in most Arabic texts. This induces both grammatical and lexical ambiguities in Arabic. The present paper describes an automatic correction of this type of errors using "Stanford Parser" with an ontology containing the rules of the Arabic language. We segment the text into sentences, then we extract the annotations of each word with the syntactic relations coming from our parser, then we treat the relations obtained with our ontology. Finally, we compare the original sentence with the corrected one in order to detect the error. The implemented system achieved a total detection of about 94%. It is concluded that the approach is clearly promising by observing the results as compared to the limited number of available Arabic grammar checkers.

Abstract-In the information systems, the query's expansion brings more benefices in the relevant documents extraction. However, the current expansion types are focused on the retrieve of the maximum of documents (reduce the silence). In Arabic, the queries are derived in many morphosemantical variants. Hence the diversity of the semantic interpretations that often creates a problem of ambiguity. Our objective is to prepare the Arabic request before its introduction to the document retrieval system. This type of preparation is based on pretreatment which makes morphological changes to the query by separating affixes of the words. Then, present all of morphosemantical derivatives as a first step to the lexical audit agent, and check the consistency between the words by the context parser. Finally we present a new method of semantic similarity based on the equivalence probability calculation between two words.