\begin{latin}
\thispagestyle{empty}
\noindent
\centerline{\textbf{\large{Abstract}}}
\\
Named Entity Recognition (NER) is one of the natural language processing which includes identifying and classifying proper names into person, location or organization. Recently, The NER task has been considered because of the remarkable influence of performance improvement on other NLP tasks such as: machine translator, information retrieval, question answering and text clustering.
\\
In this dissertation, an ANER (Arabic Named Entity Recognition) system which focuses on ancient prose is presented.  We have produced our corpus for learning and testing based on three books with three subjects including: historical, traditionary (anecdotal) and juridical. Based on ensemble learning, we trained our prediction model using Boosting method and implemented by Adaboost algorithm. POS tagging and tokenizing is applied on the corpus to overcome obstacles in Arabic language for the NER task and the results show the effectiveness of these operations. We also used context information and their properties of sequences for the feature selection. It should be noted that extracting proper names in such a morphological and rich language like Arabic is more complex than Latin.
\\
This study gathered all ANER works which used machine learning approaches. Our method is not only the first ANER but also a novel method used for NER. The model achieves an overall average F-measure value of 91.90\%. Although the model is implemented for Arabic language but it can be used in any languages.
\\

\textbf{Key words:} \
\textit{Named entity recognition, ensemble learning, Boosting method, Arabic Language}

%\textbf{Keywords:} \
%\textit{Named Entity recognition}
\end{latin}