DETERMINING ANALYSIS RECOMMENDATIONS BASED ON DATA ANALYSIS CONTEXT

• US 20150170068A1
• Filed: 06/26/2014
• Published: 06/18/2015
• Est. Priority Date: 12/17/2013
• Status: Abandoned Application

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Ganesh, Bharath R., Malviya, Rajanikant
• Application Number:
  US14/315,501
• Publication Number:
  US 20150170068A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06Q 10/063 Operations research, analysis or management

Abstract

Embodiments of the present invention disclose a computer implemented method, computer program product, and system for proposing recommendations in data analysis based on context. In one embodiment, in accordance with the present invention, the computer implemented method includes the steps of determining analytical context of an analysis step currently being performed in a data analysis, identifying a list of previously performed analysis branches that are similar to the determined analytical context, identifying a set of most similar previously performed analysis branches based on a similarity index rating associated with each previously performed analysis branch that is in an analysis tree associated with each previously performed analysis branch in the identified list, and proposing analysis recommendations for the analysis step currently being performed based on analytical context of the previously performed analysis branches in the identified set.

Claims

1. A computer implemented method for proposing recommendations in data analysis based on context, carried out by one or more processors, the method comprising the steps of:

determining analytical context of an analysis step currently being performed in a data analysis;

identifying a list of previously performed analysis branches that are similar to the determined analytical context, wherein an analysis branch is a set of analysis steps that corresponds to attributes of an analytical context;

identifying a set of most similar previously performed analysis branches based on a similarity index rating associated with each previously performed analysis branch that is in an analysis tree associated with each previously performed analysis branch in the identified list, wherein an analysis tree is a set of analysis branches that share a common analysis step; and

proposing analysis recommendations for the analysis step currently being performed based on analytical context of the previously performed analysis branches in the identified set.

2. The method in accordance with claim 1,

wherein analytical context is a set of attributes that characterizes an analysis, and

wherein the attributes utilized in the determination and definition of an analytical context include one or more of:

name, annotations, data source, concepts, measurements, hierarchies, filters, members, and analysis parameters.

3. The method in accordance with claim 1, wherein the list of previously performed analysis branches that are similar to the determined analytical context are identified utilizing semantic similarity between the determined analytical context of the analysis step currently being performed in the data analysis and previously performed analysis branches.

4. The method in accordance with claim 1, wherein the previously performed analysis branches are stored previously performed sets of steps in data analysis that include parameters utilized to perform the sets of steps in data analysis.

5. The method in accordance with claim 1, wherein the step of identifying a set of most similar previously performed analysis branches based on a similarity index rating associated with each previously performed analysis branch that is in an analysis tree associated with each previously performed analysis branch in the identified list, further comprises the step of:

identifying, for each previously performed analysis branch in the identified list, a stored similarity index associated with every other previously performed analysis branch in the analysis tree associated with the previously performed analysis branch in the identified list of previously performed analysis branches that are similar to the determined analytical context,

wherein the similarity index rating is calculated based on relative distance from an analysis branch to other analysis branches in a multi vector space and stored associated the corresponding previously performed analysis branch, and

wherein a number of previously performed analysis branches in the identified set is based on a user defined configuration.

6. The method in accordance with claim 1, wherein the step of proposing analysis recommendations for the analysis step currently being performed based on analytical context of the previously performed analysis branches in the identified set, further comprises the steps of:

determining an analysis recommendation corresponding to each of the previously performed analysis branches in the identified set by applying one or more of the attributes of the analytical context of the previously performed analysis branch to the analysis step currently being performed; and

proposing each of the determined analysis recommendations.