Last week I described the Improve Idea button that we have added to our 2013 Innovation Roadmap activity at EMC.
I'd like to describe in a bit more detail what happens when this button is pushed. The goal is to connect EMC employees with the relevant people and resources that can improve their idea before they submit it.
In a previous post I described an analytic experiment that allowed us to visualize the evolution of employee idea submission themes over a period of three years. The experiment essentially categorized employee ideas into 25 buckets using the Stanford Topic Modeling toolkit. The results, gathered in 2012, are shown below.
Each topic typically has a theme that is based on employee interest, such as "Big Data", "Cloud", or "Trust/Security".
This year, when an employee comes up with a new idea, they enter the idea into the Innovation Roadmap portal. They then have the option to push the "Improve Button". The pushing of this button causes the idea to be analyzed by the Stanford topic modeling toolkit in the context of all previous employee idea submissions. This categorizes the idea into one of the buckets.
Once the bucket has been identified, a further analysis can be run against the idea to determine "which previous idea is most closely related to yours". For example, if the employee submits an idea using terms like OpenStack technology, their idea will likely map to the "Cloud" bucket. Within this bucket a mathematical approximation will identify other documents that either (a) mention OpenStack or (b) espouse approaches that are quite similar to the inventor's idea. The more thorough the idea submission (e.g. the longer the idea), the better chance that the idea can be correlated against something similar (we have already observed that longer ideas tend to be higher quality).
For example, Topic 09 maps to Cloud Computing, and the following term evolution in that category can be seen below:
As a result of pushing the improve button, the inventor receives a list of the 10 most relevant ideas that the toolkit identifies. They can browse these ideas, view the list of inventors, and connect/collaborate as part of the idea improvement process.
What's Next?
The Improve Button has been implemented as part of an Innovation Analytics framework that goes beyond idea submission. Employees around the world can also submit other innovation activities that they are involved with (university research, conference attendance, publications, visits to start-up incubators, etc). The analytic sandbox (structured and unstructured data), and the approach of gathering innovation activities is depicted below.
Collecting this large amount of data allows our internal data scientists to create additional buckets and categories for university research, for example. The Improve Idea Button can then be augmented to analyze their idea and also refer the inventor to "most similar university research" happening within the company, reveal who is the EMC employee involved with the research, and name the professor that studies the topic.
The same approach can be used for technical conferences, visits to startups, and any innovation activity that you can think of.
The real fun starts when all of this data is analyzed together and one can begin to see which employees tend to innovate the most for a particular topic. Often times these innovators are identified as global boundary spanners that communicate their knowledge frequently. The Improve Idea button can, in theory, conclusively point to the person that is most likely to be an expert in regards to the idea being submitted.
Making helpful idea improvement suggestions relies on strong contributions to the innovation gathering framework. Variety and volume are two key parts of analytic correctness.
The more data gathered, the better suggestions can be made to improve the idea.
Steve
Twitter: @SteveTodd
EMC Fellow
