Supervised Probabilistic Latent Semantic Analysis (sPLSA) for estimating technology readiness level

摘要

The increase of global competition today encourages universities in the world including Indonesia to be able to compete with world-class universities. QS World University Rankings is an annual publication of university rankings based on six key indicators by Quacquarelli Symonds (QS). One of the indicators is academic reputation which employs survey research to evaluate the score. This paper proposes a new insight for ranking the Universities in Indonesia by making use of Supervised Probabilistic Latent Semantic Analysis (sPLSA) for extracting the level of readiness of academic journals of university staff to estimate the score of academic reputation. A corpus of keywords is developed based on Bloom Taxonomy to determine the prior level of readiness of academic journal. Supervised PLSA is employed to determine the most probable level. An academic reputation score is then computed based on the level of readiness. Lastly, we rank Indonesian Universities based on the score. To validate the result, we collect 450 abstract of academic journals from several Indonesian Universities. The results of the experiment indicate that the proposed method is promising with a distance value of 10 and similarity of 0.8 compared to the ground truth.