Home / Collection / College of Computing and Informatics / Computer Science / View Item

Search ASU IR

Trilingual Texts Feedback Gathering and Analysis for Assosa University

Eshetu Gusare, Gebreigziabher Abadi, Seiyfu Yesuf
Views: 135
Downloads: 47


When demanding to make a good judgment, we must weigh the positivity and negativity of each people feedback and consider all the alternatives. Accordingly, during decision making most of us get feedback from others and it is a natural fact that good decision can be taken on the basis of feedback of others. Before the developments of today’s technology, all the above stated facts have been experienced by asking families, neighbors, elders, friends and experts manually for making good decision. These days, people feedback is found on the Internet and despite its availability, it is unstructured and making information access challenging. To overcome this challenge, we have proposed a Trilingual Texts Feedback Gathering and Analysis for Assosa University. Feedback Analysis is the process of computationally identifying and labeling people feedback expressed in a piece of texts towards a particular topic, product, service and the like is positive, strongly positive, weakly positive, negative, strongly negative, weakly negative or neutral. As a result, this study proposes feedback analysis model on opinionated Afan Oromo, English and Amharic texts by using manually constructed rules and subjectivity lexicon of the three selected languages. The proposed model comprises of seven main key components. These are: texts preprocessing, sentiment terms detection, ambiguity detection, polarity propagation, review’s polarity weight calculation, review’s polarity classification and the developed subjectivity lexicon of the three selected languages. The developed prototype detects sentiment terms of a feedback from the developed lexicons and assigns an initial polarity weight for each of the detected sentiment terms to determine the polarity classification of the opinionated Afan Oromo, English and Amharic feedback texts. The developed lexicon of Afan Oromo, English and Amharic sentiment terms is used for recognizing and assigning an initial polarity values for each of sentiment terms detected from entered people feedback. The prototype has been developed to validate the proposed model and the algorithms designed. As a result, experiments have been done on people feedback data sets and the achieved result with these investigation data sets is very inspiring and hopeful.

Slot Deposit pulsa

View Full Item Record

Files in this Item

No file attached!

This item appears in the following Collection(s)
Search ASU IR
My Account
Date Issued
Has File(s)
  • Yes (0)
  • No (1)