Skip to main content
Planview Customer Success Center

Comparable Ideas/Idea Matching

Overview

Machine-learning capabilities allow for efficiencies in understanding ideas and concepts in Planview Spigit. As programs scale up in both user and idea numbers, the ability to identify and manage comparable ideas becomes essential. Adding to Spigit’s merger capabilities, concept matching provides another approach to how idea concepts are identified and offered for consolidation.

Key Details

Machine learning focuses upon the deeper qualitative fields of an idea. Idea owners often express multiple concepts within their ideas beyond a simple title. In normal circumstances, this means the idea's description and additional fields contain the real idea concepts and details. Most users struggle to read through whole ideas and understand fully their concepts and key points. This is a common element that contributes to challenge and idea fatigue, which can ultimately lead to issues within crowds. Against this backdrop, there is a need to better harness new capabilities to improve efficiencies for end users and administrators alike.  

COMP2.jpg

 

To identify concepts, the Spigit system uses natural language understanding and processing techniques. These capabilities leverage machine learning to read ideas, assess the words and sentences used and produce summary concepts after consulting with a learning library. The machine then produces a list of concepts for each idea that are indexed at the challenge level, which is the foundational piece for comparable idea matching.

As these concepts are more deeply rooted in the qualitative inputs from idea owners and not simple keyword matching, the action of surfacing them has powerful effects. First, idea owners are more likely to iterate and amend their ideas to make them unique. Second, it acts as a driver to the crowd to encourage them to seek out other ideas and broaden the diversity of thinking.

Once an idea is published, the system produces a list of Comparable Ideas based upon the concept matching techniques discussed above. These ideas are surfaced in the widget of the same name which can be placed upon the ViewIdea page. Clicking on the ideas title will produce the idea lightbox for that idea, allowing the user to browse the other idea from the page without the need to navigate away.

The concept matching is also embedded within the consolidate ideas feature.

FAQs

Q. Is there a threshold, a certain number of ideas, before this functionality starts to work? I've just create two ideas that have the words "electric vehicles" and "EV" in the title and description and it's not showing them as comparable. 

A. This feature extracts phrases from all text inputs in idea template fields. It does not consider idea title when extracting phrases. It matches exact phrases when finding similar ideas. If those happen to be synonyms or acronyms, then it won't find any similar ideas. Matching parts of phrases can give inaccurate results, which is why we went with complete phrases.