You have social media profiles and their associated sentiments, and you have enterprise customer profiles. Your challenge is to accurately identify and connect one with the other. The Social MDM Matching application resolves identities across the two data types (social and customer) by using high-level integration language flows.
About this task
This illustration shows the Social MDM Matching entity resolution process.
The social matching process takes social data profiles and sentiments from IBM® Accelerator for Social Data Analytics (unstructured data) and enterprise customer profiles (structured data) as input to the matching process. The matching process uses integration language flows (also called programs) to resolve entities between the two types of data. Resolving entities is another way of saying that individual records are matched together algorithmically, so that, for example, Pat Green in the social data is linked to Pat Green in the customer data.
The flows use multiple algorithms that use both integration language deterministic matching rules and probabilistic (PME) matching algorithms. The flows, run at-scale on a BigInsights™ cluster, use different subsets of available attributes, matching functions, and conflict resolution policies to link your enterprise customer profiles to external social data. The matched entity results are stored in Social MDM HBase tables.
The language flows use algorithms that are developed through extensive analytic research. There are three probabilistic algorithms (rules) and eight deterministic algorithms (rules). The flows join the results of these algorithms and PME algorithms to create the linkages between enterprise customer and social profiles.
