When searching for individuals by name or when reviewing results, you may occasionally see multiple profiles for seemingly the same individual. The reasons may be:
- It is indeed two different individuals with the same name
- Our machine learning framework has split one expert into two different profiles
We use a sophisticated proprietary machine learning framework to build up millions of searchable expert profiles in Monocl located all across the world. This includes different name parsers for individual countries (e.g. Japanese names are very different from British names, etc.), training data with thousands of known profiles to continuously improve the machine learning framework, collaboration patterns, content assignment algorithms for publications, clinical trials, grant payments, industry payments and much more.
In some cases, the quality of the underlying information is incorrect, outdated, of very low quality or simply missing all together. One good example which is easy to understand is the affiliation assignment (more information here).
Split profiles or merged profiles may also be the result of incorrect processing or failure to identify proper organizations, geographical locations and more. This is a challenge that it is faced as a constant work in progress.
We invest significant resources every day to improve our machine learning framework and measure the outcome. We encourage you to send information to us in case you encounter any split (or merged) profiles in the platform. We always review these cases and attempt to provide an update to address the issue as soon as possible.