TripAdvisor decided to translate all English-speaking user feedback into Russian

Project Tasks

TripAdvisor is the tourist website giving the chance to plan and make a trip. TripAdvisor provides recommendations of travelers and also options of planning of trips with direct references for armoring. For 2013 the websites working under the TripAdvisor brand total more than 60 million unique visitors a month and more than 75 million reviews and opinions.

Original user feedback – one of the main advantages of TripAdvisor service. But texts should be available to those who gathers in a travel on a mother tongue. Therefore for the localized websites the company does the translation of the user comments. However manual translation of tens of millions of reviews is difficult therefore the company addresses suppliers of machine translation.

The Russian tourist market with its high potential could not pass by attention of service. And for creation of the Russian-language version of the website TripAdvisor decided to translate all English-speaking user feedback into Russian. As the partner the PROMT company as the leading solution provider on machine translation from Russian and into Russian was selected.

The translation solution for TripAdvisor had to meet the following requirements:

1. The machine translation of rather high quality which is not requiring further editing for reading and understanding.

At machine translation of contents of TripAdvisor there are objective difficulties:

• The responses written by users are often very emotional, and, above all, contain a significant amount of typos, spelling, grammatical stylistic and punctuation errors that influences a translation result.
• At the beginning of the TripAdvisor project had no enough parallel texts for setup of machine translation that complicated setup of the solution under a task.

2. Automatic evaluation of translation quality

As manual assessment of all translated content of TripAdvisor is impossible, the supplier of the machine translation solution should provide the automatic mechanism of calculation for quality evaluation of translated texts.

3. Integration into production process of the website

TripAdvisor developers would like to use a cloud technology: machine translation as web service on a remote server.

For the solution of this task PROMT created and unrolled the special solution with automatic adjustment on texts of the client based on PROMT DeepHybrid technology on the platform. And special web service on the server of PROMT was developed for implementation in a technology chain of work of the website.

Project Steps

In May, 2011 trial transfer of a batch of the most demanded responses without settings for content was made. A system even in basic option showed sufficient translation quality then solution deployment with deep setup for content of the website TripAdvisor began.

In November, 2011 the solution of PROMT was completely integrated in production process of TripAdvisor and since then the user feedback is translated weekly. The translation server is placed in data center of PROMT and its work is completely serviced by specialists of PROMT.

TripAdvisor regularly provides to PROMT new linguistic data for additional setup, and the translation quality continues to improve

Project Effect

• Fast transfer of large text volumes.
• The translation quality is enough for complete understanding.
• It is much less costs for the translation of all contents, than at manual translation even of a small part of responses.
• Automatic evaluation of translation quality is used.
• The solution is integrated in production process of TripAdvisor with the minimum costs on development and support from TripAdvisor.

Responses

"The website TripAdvisor is very popular, thanks to users who write about hotels, restaurants, sights, etc. Therefore a task, crucial for business, – to make their responses available in different languages for travelers around the world. We selected PROMT for the translation from English into Russian, and it successfully works within a year. PROMT helps us to inform of experience of English-speaking travelers our Russian-speaking audience".

Lorna Whelan, the senior manager on localization, TripAdvisor

"Such large website as TripAdvisor aggregating content created by users – the ideal client for machine translation, and in particular for DeepHybrid technology. Creation of the translation system for TripAdvisor was for us a difficult and interesting challenge because of large volumes and the spontaneous nature of content. But we coped, and now tens of millions of Russian-speaking Internet users can plan trips, based on experience of other users of such popular and qualitative resource as TripAdvisor".

Epiphantseva Yulia, development director of PROMT