development of recommendation engines

bivaboti788

Since 2019, he has been engaged in the  and scoring systems for his company's service Workship, and data analysis work. Click here for a detailed profile . Here are the slides I used this time! LT.22 A story about building an environment that can run PDCA in machine learning from GIG inc. What is Workship scoring? Workship , a service currently provided by GIG , scores freelancers based on the information written in their profiles. However, there were problems in that the score distribution was difficult to spread, and it was difficult to distinguish between scores because the average score was calculated for each item.

For example, a person who scores 5 points on two items and a person who scores 0 points and a person who scores 10 points will be evaluated as having the same score even though their skills are different. So, in order to improve scoring, Mr. Sakamoto set up a function Phone Number List that increases the score depending on the amount of text in the profile. Even so, we can only judge based on quantity, not quality, so we decided to use machine learning for scoring to resolve doubts as to whether the scores are really accurate.




What has been achieved with machine learning for scoring? We asked them to introduce what they were able to accomplish while implementing the PDCA cycle for machine learning . View freelancer profiles from your browser The prediction score based on the current model is displayed and you can use it as a reference for evaluation. You can leave comments regarding evaluation criteria by linking them with user information. Anyone can annotate by sharing the evaluation method. It was a meeting where we could learn once again the importance of identifying problems and thoroughly implementing the PDCA cycle.