Introduction to Concept Parser Projects

The Classify Product from Fluree Sense comes with another powerful and unique feature: Machine Learning or AI-led Data Parsing capability. In the Semantic Object Project, which we’ve seen earlier, we were defining a Classifier.

Here, based on the Classifier’s value, we will parse Concepts further into Sub-Concepts and predict values for the same. In real world use cases, let us look at a specific scenario carrying forward from the example we had taken for Semantic Object Classification Projects.

In the Semantic Object Classification or SOC project, we had used Machine Learning to determine whether a given product was a beaker, a pipe, an electric wire, or some other such category. This could just as easily be an exercise to determine whether a patient is at risk of heart ailment or not, etc. But in this section, we’ll carry on with the first example.

In the real world, this also corresponds to the problem where companies need to structure data about that product class and present it properly rather than as unstructured or free text data such as Product Description or Short Description etc.

The below table shows an illustration of how raw data can be classified and parsed to provide customers with a structured, richer view about the Data Object.

Raw Data	Classification	Concept Parser
A 4 feet long pipe with 3” diameter of red color is available. It is made of zinc ….  A ½-inch diameter wire with 3 Ampere rating in a 30 mts bundle of pure copper 12 gauge. Grey-coloured hard glass beaker of 5 Litre capacity without spout. Certification ISO 3819. …. And more such Long Description in Data Set.	Derived Classes Pipe    Wire  Beaker	Sub-Concepts: Value after parsing: Length: 4 feet Diameter:  3 inch Color:  Red Material:  Zinc Diameter:  ½ inch Ampere Rating : 3 Gauge:  12 Length : 30 mts Material : Copper Colour : Grey Capacity: 5 Litres Spout:  No Certification : ISO 3819 Material:  Hard Glass

As we see above, the data can be parsed to a set of key-value pairs, which may differ from each category or ‘class.’

In the context of our system, these are Sub-Concepts of the associated Concept - ‘Product Description,’ linked to a classifier (‘Product Category’) for the Semantic Object: ‘Product’ in this case. The Classifier, of course, is another Concept but of a special kind. And, this is basically what Machine Learning helps us with in Concept Parser projects.