Support/Goals/Social-Feedback

Purpose

Today we are investing a big amount of time to understand what are the major issues that our users face. We want to automate and simplify part of this process by mining the data coming from feedback sources like Input.m.o and Twitter.

We will be working with the Metrics and the Community Tools team to create a reporting platform that is helpful for SUMO and the products team in general.

Project plan

The project has 4 well defined phases. This document covers the first one that is the one to be achieved in Q1.

1. Initial clustering of Input and Twitter data: Weekly reports generated by the Metrics team.
2. I18n coverage: Move from English only to Multi-lingual support.
3. Sentiment analysis of Twitter: Include signals of the positive/negative perception of each analyzed tweet. Input has this by default.
4. Off-line to online migration: Move from weekly reports to real time data gathering.

Within the first phase we will work in the following activities that will, upon completion, define the success of Phase 1 and the viability of the project.

• Define Scope: Meet with the involved parties (Metrics team, Cheng, Michelle) and define deliverables.
• Initial tests with basic dictionary: We will be performing some basic test with a basic dictionary. The goal is to start moving forward with the code (Metrics team) while the SUMO team works in a robust dictionary/taxonomy.
• Create taxonomy/dictionary: We will create the basic structure that will help the Metrics team to create clusters that are helpful for the SUMO team. The idea is to reuse part of the work that is being performed by the taxonomist.
• Standardize the access to Input data: Initially the reports will be run with ad-hoc dumps provided by the Input team. At this stage we will be working on a method that reduces the manpower and automates data collection.
• Streamline the access to Twitter data: Initially the reports will be run with ad-hoc dumps generated through the Twitter API. At this stage we will be working on a method that reduces the manpower and automates data collection.
• Embed findings in SUMO reports and iterate: The final milestone is for the SUMO team to start using the knowledge that we collect from this new source of information and document the learnings so we can improve the reports in Phase II.

How do we know it's done?

• A weekly report is generated by the Metrics team and delivered to the SUMO team and successfully used as an additional source of insight in our weekly and 6-weekly SUMO report work.

Schedule

• Define Scope: Feb 10. Depends on Metrics team, Cheng and Michelle.
• Initial tests with basic dictionary: Feb 24. Depends on Metrics Team.
• Create taxonomy/dictionary: March 2. Depends on Susan Farrell.
• Standardize the access to Input data: March 9. Depends on Metrics Team and Input Team.
• Streamline the access to Twitter data: March 9. Depends on Metrics Team.
• Embed findings in SUMO reports and iterate: March 23. Depends on Cheng and Michelle.

Blocked

No.

Notes

Next step: Define Scope. First meetings with Metrics.