High-level System Design Meeting

After meeting with Dr. Winfield, Jason and Moon proceeded to develop a high-level system design for Twine.  Essentially this is a brief overview of the guts, or inner workings, of the system that can be approached broadly and practically with sufficient time and resources.  Shown below is the diagram generated from our meeting:

Our high-level design schematic for Twine

Meeting minutes are given below.

Jason and Moon’s “Magic” Meeting – February 10, 2009 @ 3:30pm

Purpose of this meeting is to discuss exactly how the system is going to make useful information out of the tweets we get from employees.

1. From the meeting with Dr. Winfield, we decided that there should be an anonymity option when submitting a tweet – a check box perhaps for anonymous or not.

2. Identified three “models” in OOP lingo (tables in a DB)

   1. Tweets

      1. Attributes:

         1. Score

            1. Computer by:

               1. Submission addtribute (positive, negative)

               2. Parsed Text

                  1. Throw out common of unbeneficial words, such as and, or, but, it, and, etc..

                  2. Word Proximity and Pairing

                  3. Word Frequency

                  4. Punctuation

                  5. Correct misspelled words or make a guess as to what they meant

         2. Content Analysis

            1. Parsed Text

               1. Throw out common of unbeneficial words, such as and, or, but, it, and, etc..

               2. Word Proximity and Pairing

               3. Correct misspelled words or make a guess as to what they meant

               4. Determine verbs, nouns, proper names

               5. Assigned a relevancy score to an issue. Can be part of more than one issue.

   2. Issues

      1. Attributes

         1. Total tweet score

            1. Summation of both positive and negative tweet scores

         2. Positive tweet score

            1. Summation of positives

         3. Negative tweet scores

            1. Summation of negatives

         4. Controversial score

            1. Based on the variation among the scores… standard deviation

   3. User

      1. Contentment Score

         1. Summation of the user’s tweet scores

3. Issues we need to address

   1. People post single word tweets

      1. “Crap”, “Wahoo”

   2. How to handle unintentionally or intentionally high value tweets

      1. “I HATE MY FREEKING JOB!!!!###$#@%$#”

Advertisement