Thai firm translating English Internet pages

Asia Online, a Thai-based software company, has developed its own machine-translation engine, or statistical machine translation (SMT) platform, with the aim of translating English texts into Thai, to give non-English-speaking Internet users a better chance to access and benefit from vast online library resources.

Asia Online's chief executive officer and founder Dion Wiggins, was formerly vice president and research director in the Asia-Pacific region for international IT research and advisory company Gartner.

He said the power of the Internet had created an online library with an overwhelming volume of content, most of which was written in English. This creates a significant barrier for non-English-speaking people seeking to access knowledge over the Internet.

The language barrier is a particular hindrance in Asia, the region with the largest population of Internet users. It effectively creates a large part of the "digital divide".

"This big gap, on the other hand, has a big market potential for this business," Wiggins said.

The company's unique in-house SMT platform took two years for 20 Thai developers to complete. The translation system is provided to the global translation and language services industries.

Asia Online's SMT platform provides a set of techniques for automatically learning about translation from existing human translations and bilingual data, then applying that knowledge to the translation of previously unseen sentences. Instead of employing conventional linguistic rules and grammatical structures, the system uses statistics by pairing sentences in each language and learning how they are translated. Currently it supports 560 language pairs.

The company's business model is to translate as much as possible of the Internet's English-language content - involving more than 86 per cent of 31 billion available Web pages - to Asian languages free of charge, and then selling advertising to accompany the translated pages.

"Another major revenue source is generated from sales of the statistical machine translation platform to the translation industry," Wiggins said.

He said the Asian Internet market currently represented 39 per cent of global Internet users and this was expected to increase to 50 per cent by 2012. This is a huge audience for online advertising.

"From 31 billion Web pages around the world, less than 14 per cent are in Asian languages, and most of those are in Chinese, Japanese and Korean. The remaining 86 per cent of English-language content is our job," Wiggins said.

Currently, the company is working to translate Thailand's 3 million English-language Web pages into Thai, to help local Internet users to access the bulk of "useful online content".

"The firm is almost finished translating all of Wikipedia's 2.7-million Web pages from English into Thai. It has also translated open courseware from the Massachusetts Institute of Technology and Carnegie Mellon," he said.

Asia Online was recently in the global spotlight when it was awarded a Red Herring Global 100 Award, having been recognised by international technical media company Red Herring as one of the year's best global startups, with innovative and potentially disruptive technologies.

Currently, Asia Online employs more than 375 full-time and part-time staff. Most of them are Thais. The company is also in the process of being incorporated in seven other Asian countries. At present, it has business footprints in Thailand, Singapore, and Indonesia. In the near future, it plans to set up business footprints in India, the Philippines, Vietnam, Malaysia, China, Japan and Korea.

"The Thai language is well-known as one of the most difficult languages in the world. If we can successfully translate English to Thai with our machine-translation software, that means we will find it easier to translate English to other Asian languages," Wiggins said.