Browse Prior Art Database

Cognitive Association Rules Disclosure Number: IPCOM000249111D
Publication Date: 2017-Feb-07
Document File: 2 page(s) / 23K

Publishing Venue

The Prior Art Database


Increase the value of association rule models by enhancing the rules with additional information.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.


Cognitive Association Rules

Disclosed is a device to enhance regular association rules with additional information, increasing the cognitive aspect in further usage of the association rules . The idea is to not only present the association rules result , but to make use of supporting information that will either help consumers of the association rule mining results to understand the context, or to tailor the association rules model to better match the consumers.

To apply, process a given association rules models as follows :

For each discovered association rule, identify the supporting transactions in 1. the data, as well as any available accompanying data (which must not necessarily have been used for or present while building the association rules model). For each set of supporting transactions and accompanying data to an 2. association rule, post-process the accompanying data for that rule by generating e.g. univariate statistics, clustering models, etc. Possibly apply thresholds, e.g. minimum cluster size, to limit the amount of data if necessary . Add the post-processed information to each rule within the association rules 3. model, so that it is available when deploying the model later on . When deploying the model, either4.

present the post-processed information to the user to decide on his own , 1. or filter the association rules based on matching the respective user 's 2. properties to the available post-processed accompanying data using an appropriate metric.


This process can also be applied to sequential Pattern and similar models .

It is not strictly required to include the post -processed data within the model  as described in (3) - the post-processed data could also be kept separate from the model. However, for practical purposes, inclusion in the model is beneficial.

If the typical segments of consumers are known a priori , the process can be  reverted, so that the data is segmented (into intersecting portions) according to these segments prior to creating an ass...