Application of Word2Vec Algorithms to Improve Queries of Arabic Question Answering System

الموضوعات

علوم تطبيقية وتكنولوجية

الملخص

Arabic is a rich language, but there have not been many opportunities to use it in the field of scientific research. For this we suggested a way to implement the Word2Vec algorithm to improve Arabic Questions Answering System, it was tested on a set of questions written in Arabic.Word2Vec created by Google in 2013, Word2Vec use as an entry a large or small text and then produce a vector space usually of several hundred dimensions.Word2vec has been used in several domains, from these domains Question answering system with different language. But we noted that these works are not very applied to our mother language "Arabic language".We suggested a way for implement Word2Vec algorithm to improve Arabic Question answering system, it was tested on a set of questions written in Arabic.Our method is to improve questions in Arabic language, then apply the Word2Vec algorithm with its models, on this questions after doing a preprocessing to extract similar words in context or meaning, then try to find the answers to the reformulated questions via Google API. Only the relevant answers are kept for each question.Other times, we take the initial questions without preprocessing (which means in the original state) and try to find the answers also via Google API, and we keep just the relevant answers.From the results we get after doing the two parts, we compare between the two results (initial and reformulated), to see if there is an improvement or not according to the percentage.

روابط وملفات

التعريف والنوع

رقم الوثيقة

f5ccc883-9646-467b-90e1-a851b629bb6a

رقم العقد

0

نوع الوسائط

Crawler

نوع المحتوى

الرسائل العلمية

صيغة المصدر

ماستر (LMD)

نوع الملف

pdf text

أسماء الملفات

2260507_1.pdf

بيانات النشر

ألقاب المؤلفين

[{"name_ar":"Ahmed Cherif Mazari","title_ar":"اشراف","title_en":"Supervision"},{"name_ar":"Amira Irki","title_ar":"اعداد","title_en":"Preparation"},{"name_ar":"Mohamed Ramzi Hachemi","title_ar":"اعداد","title_en":"Preparation"}]

اللغة

English

المصدر والدورية

اسم المصدر

Application of Word2Vec Algorithms to Improve Queries of Arabic Question Answering System

المحتوى والصفحات

عدد الصفحات

0

كلمات الباحثين

Arabic language Word2Vec Question Answering System Similarity Deep Learning

إشراف وإعداد

الإشراف

Ahmed Cherif Mazari

الإعداد

Amira Irki, Mohamed Ramzi Hachemi

الاقتباسات الببليوغرافية