1. Insights
  2. AI Data
  3. Case Study

Improving search relevance for a fast-growing unicorn

Learn how TELUS International classified over 200,000 search queries for a leading online travel company which provides a one-stop platform for a range of ticketing services including flights, accommodation and attractions.

thumbnail image
  • Share on X
  • Share on Facebook
  • Share via email

+200K
text strings

50+
professional contributors

6000+
hours of work completed

The challenge

The search function of a website is of critical importance. An effective search engine quickly and effortlessly directs customers to the most relevant products based on their query. As a result, search accuracy can have a huge impact on customer retention and sales volume.

Our client, a Southeast Asian online travel company, provides a one-stop platform for a range of ticketing services, including flights, accommodation and attractions. With an expanding list of 19 core product offerings, improving search capabilities was key to their continued growth across multiple platforms. To do this, our client aimed to build a universal search function, which would enable customers to browse the full range of their products from a single search bar.

To complete this complex machine learning project, they required a data annotation partner who could classify thousands of search queries according to an extensive system of product and sub-product categories. This partner also needed to have a specialized skill set annotating similar text strings with their linguistic meaning.

The TELUS International solution

Our ability to perform multiple types of data annotation in both English and Indonesian was instrumental in the client’s decision to partner with us.

Key differentiating factors of our data annotation solution included:

Guideline preparation

Working closely with our client, we developed extensive guidelines for our contributors in both English and Indonesian.

Text classification specialists

We carefully selected, tested and awarded qualifications to a small group of contributors who are part of our larger AI Community. This heavily-vetted subset consisted of 50 native speakers of both languages who annotated the client’s data using our custom text classification workbench.

Quality evaluation

We thoroughly reviewed our contributor output to ensure that it met the client’s high standards for consistency and quality. We also developed a script to convert the classification output into IOB2 format.

The results

Using data annotated by TELUS International, the client launched a search engine that allows users to effortlessly search 76 unique product combinations with a single click. This machine learning algorithm continues to improve ease-of-use for both their web and app users.


Check out our solutions

Ensure that your model performs at the highest level of accuracy and precision against real world data.

Learn more