Dynamic natural language processing

A computer program product to support natural language processing, the computer program product comprising a computer readable storage medium having program code embodied therewith, the program code executable by a processor to:

• analyze received natural language (NL) input, including identify a lexical answer type (LAT);
• identify a knowledge graph (KG) related to the LAT, and leverage the identified KG to extract a first concept in the KG related to the LAT and to extract a second concept in the KG related to the identified first concept;
• create two or more clusters, the two or more clusters comprising a first cluster and a second cluster, the first cluster having a first combination of two qualifiers selected from the LAT and the identified first and second concepts, the second cluster having a second combination of two qualifiers selected from the LAT and the identified first and second concepts, the first combination differing from the second combination, and populate each of the created first and second clusters with one or more documents related to the first and second combinations, respectively;
• conduct intra-cluster assessments of the first cluster and the second cluster based on relevancy to the received input to identify a most relevant first document of the first cluster and a most relevant second document of the second cluster, respectively;
• conduct an inter-cluster assessment based on relevancy to the received input for cross-cluster ranking of at least the first and second clusters;
• and
• identify a MLM corresponding to the identified KG, selectively augment the MLM with the LAT, the first concept, the second concept, and a corresponding relationship responsive to the inter-cluster assessment.

A method comprising:

• analyzing receiving natural language (NL) input, including identifying a lexical answer type (LAT);
• identifying a knowledge graph (KG) related to the LAT, and leveraging the identified KG to extract a first concept in the KG related to the LAT and a second concept in the KG related to the identified first concept;
• creating two or more clusters, the two or more clusters comprising a first cluster and a second cluster, the first cluster having a first combination of two qualifiers selected from the LAT and the identified first and second concepts, the second cluster having a second combination of two qualifiers selected from the LAT and the identified first and second concepts, the first combination differing from the second combination, and populating each of the created first and second clusters with one or more documents related to the first and second combinations, respectively;
• conducting intra-cluster assessments of the first cluster and the second cluster based on relevancy to the received input to identify a most relevant first document of the first cluster and a most relevant second document of the second cluster, respectively;
• conducting an inter-cluster assessment responsive according to relevancy to the received input for cross-cluster ranking of at least the first and second clusters;
• and
• identifying a MLM corresponding to the identified KG, selectively augmenting the MLM with the LAT, the first concept, the second concept, and a corresponding relationship responsive to the inter-cluster assessment.