Browse Prior Art Database

Method for discovering a similar past project or work item by analysing unstructured data Disclosure Number: IPCOM000245701D
Publication Date: 2016-Mar-31
Document File: 2 page(s) / 55K

Publishing Venue

The Prior Art Database


The invention describes a method for finding a similar project to one whose duration has to be estimated. This is the basis for analogy based estimation. The accuracy of the results that this method of estimation provides depends on the how similar the base project is with the one to be estimated. The invention finds a similar project by analysing the data associated with the current project (typically a summary, some mockups, a list of use cases) and a repository of past projects. A variety of analytics techniques are used to find similar projects: keywords are identified and expanded with similar concepts. Mockups are analysed and concepts get extracted and matched. The list of similar projects is further validated by computing a similarity score. The result of the invention is a list of past projects that are similar with the current project, along with confidence scores for the similarities. These projects can be used as a basis for analogy based estimation.

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

Page 01 of 2

Method for discovering a similar past project or work item by analysing unstructured data

Analogy Based Software Estimation is a well known technique for estimating the size of projects, stories or tasks. The technique is based on the principle that actual values achieved within the organization in an earlier and similar project are better indicators and predict the future project performance much better than an estimate developed afresh from scratch.

    A high quality of an estimation is directly linked with the quality of its starting point and it is difficult to find a successful similar past project.

Although software for planning and software management allows to search for a similar work item in the repository, this has some limitation to the searched sources.

    An incomplete data source can result in incorrect decisions made based on limited data. The solution herein proposes a new idea of using the complete data sources to discover the duplicated or similar work items for decisions on work item life cycle management.

    The solution herein will be expressed using the finding of a good analogue to get an estimation for a story or triage for a defect or PMR. Something similar can be applied for projects and tasks.

    An initial story that needs to be estimated, typically has a title, sometimes a short description, in other cases a use case document, or an email thread where the idea was mentioned and possibly some mockups.

    The solution uses these starting points to examine against its repository of known completed stories to come up with good candidates for matching. The key here is how the input is used for searching.

    Below are the main steps of the solution 1. Identify the keywords in an item summary 2. Expand the list of keywords using concepts insights and expansion and visual recognition techniques 3. Use image searches if any screen shots are present 4. Find the potential matches?

5. Validate the potential matches and display with a confidence score
The solution comprises
- the combination of simple text matching algorithms with more complex analytics like concept insights and expansion and visual recognition for an extended list of matches

    - use of image searches to find matching stories
- validating the results using a different technique to reduce the likelihood of false positives.

    The solution herein is to go from the details of the story to the keywords, the extended keywords, then search for similar stories using the extended keywords and then validate the results using a different met...