Please use this identifier to cite or link to this item:
Title: Abiguous Arabic Entities for Web Search Results Disambiguation
Other Titles: Arabic Search Reasults Disambiguation
Authors: Yahya, Adnan
Hamad, Ruba
Sandouka, Rand
Keywords: Arabic NLP
Fake Video Captions
Issue Date: 20-Jun-2019
Publisher: Birzeit University
Citation: Ruba Hamad, Rand Sandouka, Adnan Yahya; Fake Video Arabic Captions Dataset
Series/Report no.: Arabic NLP Dataset;1
Abstract: Machines too like humans are capable of learning once they see relevant data. But where they vary from humans is the amount of data they need to learn from. You need to feed your machines with enough data in order for them to do anything useful for you. The data-set in some cases must be very large to enable sufficient learning for the model to be generated; therefore, the experiment will be transformed to a data collection task.
Description: 1. First of all, we have chosen 20 Arabic words that have at least two meanings, which are called ambiguous entities, e.g., لحام، سهم، عشاء. 2. We have collected 50 document links for each word with its different meanings using Bing search engine. 3. We have chosen 8 judges, with 4 judges for the first ten words and 4 judges for the rest of the words. 4. Then we distributed forms (shown in A1 & A2) among judges to fill them. 5. Lastly, we calculated Kappa.
Appears in Collections:6. BZU Dataset Collection

Files in This Item:
File Description SizeFormat 
RubaHamadRandSandoukaNLP_MajorAssignment.rar553.38 kBUnknownView/Open    Request a copy

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.