Turning the tables into texts. Generating hotel descriptions automatically

In the highly competitive travel industry, demand for timely, accurate and informative textual overviews is on the rise. Pages with strong content increase user engagement and PageRank. At HolidayCheck, we have established that there is user demand for textual descriptions of travel entities, such as hotels. As the quantity of required texts increase, manual production becomes prohibitively pricey and the large body of text is next to impossible to update and maintain. In this talk, I will describe how data-to-text NLG techniques have been applied to generate unique descriptions for more than 200 000 entities rapidly. This approach makes it possible to update and regenerate the descriptions easily. The presentation will introduce the main stages of knowledge acquisition as well as the major steps of the actual text production in Python

Papp Kornélia
Natural Language Processing Engineer, HolidayCheck

Kornélia Papp is a Natural Language Processing Engineer in the Content team at HolidayCheck AG Switzerland. She currently focuses on content generation and enrichment in NLP-driven products. She previously worked on several projects developing language technology applications in the area of data analytics and machine translation. Before joining HolidayCheck, she was working in the Speech Technology Industry developing synthetic voices. Kornélia holds a PhD in cognitive linguistics and has more than 10 years’ experience in language technology. She is an active member of the Python Meetup Team in Zurich, Switzerland.