Hi Rasha

What kind of program do you mean?

If you want to develop an opinion mining program that detects sentiment of tweets in Arabic for example, then yes I would say you most probably need an Arabic lexicon.

There is a paper by Badaro et. al (2015) describing Arabic lexicon for sentiment.

The lexicon is downloadable from this website: www.oma-project.com/ArSenL

Also, you might want to check  the paper for more details (A Large Scale Arabic Sentiment Lexicon for Arabic Opinion Mining) .

Best

Hi Rasha,

 In any language, (in your case, Arabic), you should study and work with not only the lexicons, but also with the morphology ( how to construct different forms of lexicons), as well as the syntax. Although, for some application you may only need the lexicons, that you can follow the Abdulaziz Aamri answer, but for sentiment analysis, I believe you need more work with the syntax too.  

 There are many papers about sentiment analysis in Arabic, I have reviewed some of them myself for the Journals. but they are beyond that just building a lexicon dictionary. 

I hope it helps.

Mostafa

Thank you very much for Abdulaziz and mustafa.I am working in project that summarizing arabic documents.The problem that I have is how to make the program understand arabic rather than english in the simplest way,and I need to make it work so quickly ,cause I have not much time,may be less than month.When I asked friends that they know about NLP ,they told me that I should create annotation for every words by using brat ,but  I think that may be take alot of time,especially that I know nothing about the annotation or the brat,and I don't know also how I can merge it with working project.Is it possible to make project that summarize arabic text without annotation?if the answer is no,what is the simplest way to did it?

Hi Rasha,

 Brat is just used for annotation and it is usually used to prepare data (in their format) for training machine learning techniques. It means their output has to be transferred to the required format for any machine learning techniques. 

 Your problem, as I found is more than that only annotation. There are special tools for any language like English or Arabic for Sentence/Token segmentation, POS tagging, NER, Topic Discovery, etc. I do not know the structure (or tools/libraries) that your current system for English have, but if you want to use it to summarize the Arabic documents, you need to map your system using Arabic-related tools (NLP tools) for example, tokenizer for arabic, POS tagger for Arabic, etc. 

If you explain more about the current status of your project and, I may be able to help you more. I have some experiences in English document summarization using computational linguistics (using various NLP tools, such as POS tagger, NER tagger, Chunking, Topic Discovery, Sentence ranking(scoring), ...) and I also a little bit familiar with the right-to-left languages like Arabic or Persian. 

 I hope it helps. 

Mostafa

Alsalam Mostafa,

Thank you for quick response.I am using algorithm that based on Graph-based Lexical Centrality ,we implement the algorithm ,we did the limmitizer and  and used some equations,but we do not know how should we can make it understand arabic text,We should do it in preprocessing,but what is the best way to did it.Thank you

Hi Rasha,

 I guess you mean Lemmatizer! I am not familiar with the Graph-based lexical Centrality and how it is used for summarization. But, as the answer to your question and based on what I understood from so-far conversation, I believe you need arabic Tokenizer and Lemmatizer at least. Then, I guess you need to customize the question based on the semantics in Arabic. 

 I am not sure, but I guess there should be some tokenizer and lemmatizer for Arabic. 

I hope it helps. 

Mostafa

Salam mustafa,

I did not mean the limmitizer ,we did the basic operations that the algorithm require like use idf,remove stop words,and the limmitization,but what we did is to applay the algorithm,what we need is when our project read arabic text ,we need to understand words,and give them values,based on there importance,that means when the word is more important there must be more possibility to be included in the summary.There are many important words for every field,who the project can understand the text and know that this text relate to specific filed and that this word is important at this field.Thank you.