Information Technology for Finding Answers to Questions from Open Web Resources

The purpose of this work is to develop a question answering system for the questions related to the highest degree of comparison based on open web resources. Examples of such questions are: "What are the most popular cars 2023?", "What is the best novel ever?". The system is implemented in Python. Additional libraries were used like Selenium for web scraping and NLTK for keywords extraction. The result of the system is ranked list of items with number of occurrences. The ranking is based on the popularity of item, or more specifically on the number of web resources in which the particular item appeared. It differs from the classical QA systems because they usually answering unequivocal questions. In contrast, implemented system aggregates opinions from different resources to answer subjective questions. There are no datasets for our specific task, so the system was validated on the manually created sample questions. Considering the above limitation system accuracy expressed in F1 score is close to 0.8. In future the system can a website that aggregates data from many web resources designed to rank items from different domains.