{"id":221995,"date":"2024-08-16T15:27:27","date_gmt":"2024-08-16T07:27:27","guid":{"rendered":"https:\/\/www.grab.com\/sg\/?post_type=editorial&p=221995"},"modified":"2024-09-16T15:25:54","modified_gmt":"2024-09-16T07:25:54","slug":"automation-ai-menu-translations","status":"publish","type":"editorial","link":"https:\/\/www.grab.com\/sg\/inside-grab\/stories\/automation-ai-menu-translations\/","title":{"rendered":"With automated menu translations, GrabFood merchants reach more consumers"},"content":{"rendered":"\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t
\n\t\t\t
\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t
\n\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t

Ordering food in a restaurant in another country can be a struggle, especially if you\u2019re not familiar with the local language. It\u2019s the same when you\u2019re ordering on the GrabFood app too.<\/p>

We began to help our merchant-partners translate their menus into various languages so that they could reach a wider segment of consumers, such as travellers and expats who don\u2019t speak the local language. For consumers, seeing the menu in their preferred language makes ordering much easier.\u00a0<\/span><\/p>

However, with so many merchants on the platform, manually translating each and every item would be time consuming. The solution? Automating the process.\u00a0<\/span><\/p>

In Southeast Asia, translations have to consider many local languages and their nuances. To do so, we tap into deep learning models with our own Southeast Asia-specific data sets. By training our models, we were able to automate the previously tedious task of menu translations.<\/span><\/p>

(Read more: How AI helps our search bar seemingly read minds<\/a>)<\/strong><\/p>

Today, over 90 per cent of our merchant-listings in Thailand and Indonesia are available in English and Chinese languages; and about 90 per cent of listings in Singapore are available in Chinese. We\u2019re also piloting English translated menus across five cities in Vietnam and are working to expand menu translations to more languages.\u00a0<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\"\"\t\t\t\t\t\t\t\t\t\t\t
Users can tap on the translation icon to instantly view the menu items in their preferred language.<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t
How automated menu translation works<\/h5>

We started off by building a model from scratch using open source subtitles extracted from movies. However, we found that the translation accuracy was low because the data couldn\u2019t capture the local nuances in Southeast Asian cuisine.\u00a0<\/p>

To localise our menu translation model, we trained it with our own dataset. We labelled the names of top transacted food items. This allowed us to train our system to understand and analyse different dish names, ingredients, and descriptions, which led to more accurate translations.\u00a0<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t

Further improving the model, our team manually collected inaccurate translations to fix them. Over time, we found that manually annotating the spread of data is impractical, if not near impossible, given the vast number of food items listed on our platform.\u00a0<\/p>

So we used the model to automatically assess the accuracy and reliability of each translation by assigning confidence scores.\u00a0<\/p>

When a translated food item receives a high confidence score, it indicates a high level of certainty in the accuracy of the translation. On the other hand, lower confidence scores may suggest potential inaccuracies or ambiguities in the translated text, helping us to sieve out poorly translated items more efficiently.\u00a0<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t

\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\"\"\t\t\t\t\t\t\t\t\t\t\t
Our in-house translation model is more accurate than off-the-shelf translation models<\/figcaption>\n\t\t\t\t\t\t\t\t\t\t<\/figure>\n\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t
Making menus readable for everyone<\/h5>

Users can simply tap on the translation icon to instantly view the menu items in their preferred language. This feature eliminates the need to rely on guesswork or seek assistance from others when trying to order food in a different language.<\/p>

Grab’s menu translation feature is constantly evolving and aims to serve users from all over the world by expanding the number of languages supported and enhancing translation accuracy using machine learning. We hope it will bring a better experience to our users.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t

\n\t\t\t\t\t\t
\n\t\t\t\t\t
\n\t\t\t
\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<\/div>\n\t\t","protected":false},"parent":180237,"menu_order":0,"template":"grab21-default","acf":[],"_links":{"self":[{"href":"https:\/\/www.grab.com\/sg\/wp-json\/wp\/v2\/editorial\/221995"}],"collection":[{"href":"https:\/\/www.grab.com\/sg\/wp-json\/wp\/v2\/editorial"}],"about":[{"href":"https:\/\/www.grab.com\/sg\/wp-json\/wp\/v2\/types\/editorial"}],"version-history":[{"count":25,"href":"https:\/\/www.grab.com\/sg\/wp-json\/wp\/v2\/editorial\/221995\/revisions"}],"predecessor-version":[{"id":224194,"href":"https:\/\/www.grab.com\/sg\/wp-json\/wp\/v2\/editorial\/221995\/revisions\/224194"}],"up":[{"embeddable":true,"href":"https:\/\/www.grab.com\/sg\/wp-json\/wp\/v2\/editorial\/180237"}],"wp:attachment":[{"href":"https:\/\/www.grab.com\/sg\/wp-json\/wp\/v2\/media?parent=221995"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}