Home About Login Current Archives Announcements Editorial Board
Submit Now For Authors Call for Submissions Statistics Contact
Home > Archives > Volume 20, No 8 (2022) > Article

DOI: 10.14704/nq.2022.20.8.NQ44796

Automatic Extraction of Semantic Relation among Verses of Quran's

Haneen alaa Alhaidari, Ahmed Hussein Aliwy


Many natural language processing applications such information extraction and sentiment analysis need extraction of semantic similarity and semantic relations among documents, passages, sentences and phrases. Extracting semantic similarity is big problem in rich language such as Arabic languge and this problem become more difficult in case of classical Arabic where many different texts refere to same meaning. One of the best example of classical Arabic is The holy Qur’an; the most important constitution for 1.8 billion Muslims on the world. All Muslims judgments are induced from this book with assistance of other resources. These judgments cannot be accurate without complete knowledge in Qur’an and the semantic relationships among its verses. Manual detection of semantic relationship of verses is hard task because it contains 6236 verses. In this paper, we developed a model for detecting semantic similarity among the verses based on word embedding. Three datasets were used; Qur’an corpus as a raw text and as assistant to produce word embedding, AraVec as source for producing word embedding, and Qursim for evaluation of relationships among the verses. Word embedding were estimated from Qur’an corpus and then merged into these produced from AraVec dataset. The proposed system was implemented using five similarity/distance equations; Euclidean distance, Manhattan distance, Minkowski distance, correlation, and cosine Similarity. Hence, the accuracy, precision, recall and f-measure were calculated to assess our model's performance. The obtained result show that accuracy for cosine scale is the highest, where the results for cosine scale were 0.81, 0.99, 0.89, and 0.88 for precision, recall, f-measure and accuracy respectively. These results ensure the effectiveness of our model for Qur’an corpus and hence ut can be used for classical Arabic


Semantic relation,word embedding,sentence embedding

Full Text