Structured and unstructured data capture


SDC has been offered as a solution to reducing the costs of data entry by converting the Market Reform Contract into an ACORD Global Placing Message by use of OCR technology and as such is based on a paper based rendition of the MRC. Within PPL, work is underway to create a GPM from a Word version of the MRC that is not dependant on OCR and thus not dependant on paper. This would appear to be a step in the right direction in digitising the contract data in a structured form when the basis of most contract data during the placing process is documents and maybe spreadsheets. London is of course fortunate in that it has a uniform template for the unstructured data in the MRC.

The recent information coming from LIIBA that data can be made available to carriers pre-bind will add a lot more weight to the SDC business case in that data can be made available to carriers at the quote stage, which is when many carriers enter the contract data for the first time and then they develop that contract record as the placing process continues. SDC post bind ‘missed the boat’ for many carriers thus removing the data entry benefit.

Other territories are not so fortunate in having an MRC style template and in fact the variation of submission and slip formats is extensive. The problem is the same as that facing those who receive coverholder bordereau and that is they have no control on how the sender sends the data.

The situation in the mid market segment is particularly demanding given the number of brokers and thus the potential for variability in submission formats, but the high volume of such submissions. This is an area where we can apply some innovative thinking and an innovative technical solution.

Digitising the data from Word, pdf, emails etc in its raw form is not the challenge – there are solutions around that can do that. The challenge is how to convert the unstructured data into structured data. If this holy grail can be achieved then the beneficial digital transformation possibilities are significant.

One way of achieving this is to apply Natural Language Processing concepts. There are a number of development languages that have NLP capabilities, but the key is in developing the ontology that the NLP processing relies on.

I see the application of NLP to unstructured data as a good investment of innovation budgets considering the potential benefits.