Lets say we have an English to French translation task in a company and there are 100s of workers who are proficient in doing this task but each worker has its own unique attributes which allow them to do certain translations in a better way over others eg: A translator who is a doctor would do better translations of medical documents. Consider 1000s of documents for translation received per day and the turnaround time is 1 day per document. So there is a two fold problem

1)Allocating the tasks to available translators using a queuing model or any other efficient allocation mechanism.

2)Learning the worker model based on his past performance like correctness scores, static parameters like skill set, qualifications, experience etc using a machine learning system

So is there an setup / system / model which solves both of the above problems in one cohesive system. From my reading this problem calls for application of the assignment problem , user modelling, queuing and machine learning optimization. I am searching for existing models or frameworks which integrate all of this.

  • $\begingroup$ I'm not aware of ready-made "cohesive" recipes (they probably can be found in industrial analytics easier than in statistics). Task 1 to match workers and works could be done via Hungarian algorithm with cost (or gain) as some propensity towards a specific type of document. Task 2 could fashioned in regression or even time series with predictors. $\endgroup$ – ttnphns Dec 10 '15 at 17:40
  • $\begingroup$ @ttnphns Any framework which integrates both? $\endgroup$ – stats101 Dec 11 '15 at 6:05
  • $\begingroup$ Any current state of worker's ability (2) could be incorportated into a coefficient of his current propensity towards task(s) in (1). You are free to select or invent reasonable formula to compute the propensities (values of the gain matrix). $\endgroup$ – ttnphns Dec 11 '15 at 10:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.