How Betterhalf Uses AI To Match Users To Their Compatible Partners?

Betterhalf

Artificial Intelligence, commonly known as AI, is one of those terms that has become more mainstream with each day. Such is its popularity that we can find it in our smallest of life aspects. In our country, we can also attribute the never-seen-before internet penetration to the emergence of AI in our life. Other than this, we can also agree that life has indeed become much more convenient for all of us since the mass press coverage of AI in the early 2000s. Technology and innovation have been helping the world become faster than the previous day. From self-driving cars to e-commerce, facial recognition, music platforms, and so forth, we have recommendations for everything in today's time.

It would not be a big thing that you're reading this article because of the recommendation of one algorithm. However, despite so much advancement in technology, there is one industry that has remained untouched by it for potentially 1,000 years, or maybe more - the partner-search industry.

Have you ever wondered why? What is the reason that we haven’t seen the involvement of AI in finding our Betterhalf?

Well, we discovered that while the industry sees traditional matchmaking through one model or the other in the terms of matching people — which, ideally should be the last step of the partner search journey — they don’t talk about the entire search process, which is way more volatile, uncertain and frustrating and anxiety-driven.

At Betterhalf.ai, we thought of breaking this pattern. Here, we aspire to transform an uncertain journey to certain, timely, and delightful for 500M people globally through an AI-based partner prediction engine. Our engine starts learning about a user’s personality as soon as s/he starts the onboarding process through the use of AI in five different stages. We will take you through each of these five steps so that you can understand better.

During registration of users

At the first stage, we capture users’ personalities in six different relationship personality dimensions - Emotional, Social, Intellectual, Physical, Relationship, and Values. Here, we ask them a series of 16 Likert-type questions. With the answers given by users, we’re able to estimate one’s initial personality and background information with reliable and impeccable accuracy. Besides this, we use in-product gamification, pre-match and post-match activities of the user, and their feedback about other users to get more information about the preferences to find them their Betterhalf.

During pre-chat/conversation stage

Once users are on board and they start to interact with the product, we capture their behavioral information such as click-map, scroll-map, time spent on different sections of their matches’ profile, etc., to learn more about them. Let’s understand this in more simple terms - Suppose a user has visited ten matches, and 5 of them have mentioned that they like to travel. Now if the user spends more time with these profiles, our engine will automatically learn that this particular user has an interest in matches who like traveling.

Similarly, if another user spends comparatively more time on those profiles who are in the healthcare profession and ignore other profiles, then this behavior will imply that the user is more inclined towards marrying a healthcare professional. So, that’s how our AI engine learns about users and takes care of their preferences.

During product gamification

This stage is a crucial part of the whole journey. Through product gamification, we try to capture additional personality information from users on our platform. And we do it in a gamified way where we ask more personality-based questions from users to learn more about their personality, what kind of partner they like to have in their life, etc.

Once users spend more time on the product, we get this additional data over time, and it helps us rectify any personal biases cropping in users’ minds due to some bad experience that they might have faced that day. We don’t ask all the questions at once because asking questions over a period also helps us capture the information in different stages of the journey and thereby, helping us evaluate the exact personality of a user. For your information, we use an AI-based algorithm to probabilistically update and correct initial personality representation. So, you don’t need to worry about any biases and preferences mismatches.

During post-chat stage

This stage comes after a chat with their match. At this stage, we take timely private star-rating feedback about the user from their matches once they have interacted or chatted with them. This feedback covers a wide range of topics like the authenticity of the profile, intent to settle down, timely & quick response, compatibility with a match to various others likes or dislikes about their matches. Here, we again take the help of AI to refine the user’s profile and personality representation based on this feedback from other users. This process helps our AI engine create a more accurate version of ‘user’ and their personality'. This is the most important part to help us provide compatible matches to you.  

Removal of any outliers/biased data using Machine Learning (ML):

The final step in the partner-search journey is to remove and rectify any outlier data. And how do we do it? By using an ML algorithm. Let’s suppose if, on a short-tempered dimension, 99.9% of users rate themselves between 1 to 4, and if there’s a user who has marked himself as 7, we will apply ML-based rectification to improve this data.

Our Compatibility Estimation and Matching formulae, derived from research work going on in Cambridge University [1], [2], [3] and [4], [5], [6], [7], [8], [9], and [10], utilize this personality representation to predict best compatible matches for users.

At Betterhalf, we use a person’s basic partner preferences such as age range, height, caste, religion, location, education, salary, etc., to filter and rank those matches. This is exactly how users receive matches. And as they interact with more and more users, their matches improve over time, thereby leading them to find a compatible Betterhalf sooner than they expected. That’s how we are trying to help people find their marriage partners.

References:

  1. Donnellan, M.B., Oswald, F.L., Baird, B.M., & Lucas, R.E. (2006). The mini-IPIP scales: Tiny-yet-effective measures of the Big Five factors of personality. Psychological Assessment, 18, 192–203
  2. https://discovermyprofile.com/personality.html
  3. https://applymagicsauce.com/demo.html
  4. Daniel Nettle. Personality: What Makes You the Way You Are (Oxford Landmark Science), ISBN: 9780199211432
  5. https://research.peoplematching.org/
  6. https://openpsychometrics.org/_rawdata/
  7. Henley, N.; Meng, K.; O’Brien, D.; McCarthy, W.; Sockloskie, R. (1998). “Developing a Scale to Measure the Diversity of Feminist Attitudes”. Psychology of Women Quarterly, 22(2), 317–348
  8. Hirschfeld, Gerrit, Ruth von Brachel, and Meinald T. Thielsch. “Selecting items for Big Five questionnaires: At what sample size do factor loadings stabilize?.” Journal of Research in Personality (2014)
  9. http://ipip.ori.org/
register now