Evaluation of Automatic Text Visualization Systems: A Case Study
| Autor: | Karimullah Shaik, Priyanka Jain, Hemant Darbari, Virendrakumar C. Bhavsar, R. P. Bhavsar, Ajai Kumar, B. V. Pawar |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Hindi
Design evaluation business.industry Computer science media_common.quotation_subject Interactive 3d Fidelity Intelligibility (communication) computer.software_genre language.human_language Visualization language Online evaluation Artificial intelligence business computer Natural language processing Natural language media_common |
| Zdroj: | Advances in Intelligent Systems and Computing ISBN: 9789811533822 |
| DOI: | 10.1007/978-981-15-3383-9_3 |
| Popis: | We have developed an automatic text visualization (ATV) system, named Preksha, that takes natural language text in Hindi as input and produces a run-time interactive 3D scene based on it. Preksha is the only ATV system which deals with complex processing of morphologically rich input in Hindi, a language of free-word-order nature. Its design and approach make Preksha extendible to other Indian languages. In this paper, we present challenges for evaluation of an ATV system and propose a subjective evaluation methodology. This evaluation design includes intelligibility, fidelity and complexity aspects of the scenes generated. Subsequently, Preksha is evaluated by using a total of 10,220 user responses through an online evaluation survey. The results show that Preksha is able to generate scenes with very high levels of intelligibility and fidelity. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|