Difference between revisions of "Kauffman Incubator Project"

From edegan.com
Jump to navigation Jump to search
Line 24: Line 24:
 
# Classify entrepreneurship ecosystem organizations, including high-growth technology incubators, startups, and venture capitalists based on a short textual description;  
 
# Classify entrepreneurship ecosystem organizations, including high-growth technology incubators, startups, and venture capitalists based on a short textual description;  
 
# Identify the client listing page on an incubator's website;  
 
# Identify the client listing page on an incubator's website;  
# Automate the extraction of information about startups from an incubator's client listing page; and
+
# Automate the extraction of information about startups from an incubator's client listing page;
 
# Make this system available to the research community as opensource software.  
 
# Make this system available to the research community as opensource software.  
  

Revision as of 11:51, 5 March 2019

Kauffman Incubator Project
EMKF Stacked RGB brighter Blue crop.png
Project Information
Principal Investigator Ed Egan
Academic Institution Georgetown University
Grant Cohort UMM
Copyright © 2019 edegan.com. All Rights Reserved.

Introduction

The Kauffman Incubator Project will advance the capacity of researchers and policymakers to measure the characteristics and performance of entrepreneurship ecosystem institutions. To this end, we are creating a set of tools to automate the identification and classification of ecosystem organizations, and the extraction of data on startup firms from their websites.

Specifically, our goals are to develop a system to:

  1. Classify entrepreneurship ecosystem organizations, including high-growth technology incubators, startups, and venture capitalists based on a short textual description;
  2. Identify the client listing page on an incubator's website;
  3. Automate the extraction of information about startups from an incubator's client listing page;
  4. Make this system available to the research community as opensource software.

We are using modern machine learning techniques in our tools, and we expect that the third element will require new computer science.

Expected Outputs

By March 2019

  1. Determine at least 4 primary data sources, or secure licenses to extract ‘seed data’ from these sources, as measured by program records.
  2. Have a working prototype of an automated classifier to distinguish between incubators and other entities described in seed data, as measured by program records.
  3. Collect data in at least 5 ecosystems, as measured by availability of a dataset.
  4. Develop a protocol for the tool to extract client company identity information from incubator websites, as measured by program records.

By June 2019

  1. Have a working prototype of a tool to identify client company listings from incubator websites, as measured by program records.
  2. Upload the collected data to GitHub, Dataverse, or other publicly accessible web platform for use by a set of academics, as measured by program records.
  3. Produce a summary on the open development process for the prototype as measured by program materials.

Expected Outcomes

By June 2019

  1. At least 3 improvements to the measurement system will be made as a result of collaboration with other researchers, as measured by feedback from collaborators.
  2. At least 15 researchers external to the UMM cohort will have contacted the project for additional information and/or collaboration, as measured by correspondence.
  3. At least 5 outside users will have used the open source tool, as measured by records of use and postings of the data.
  4. There will be at least 50 views of the online documentation of the development process, as measured by web analytics.
  5. The data posted on the public forum will have at least 50 views/downloads of the uploaded data, as measured by web analytics.
  6. The seed data will have at least a 70% baseline accuracy and coverage of incubators compared to results from hand collected data on 5 ecosystems, as measured by the data analysis.