WEEK-11

WEEK 11:

Day 1: I started the week by removing the attributes, that was done the previous week and adding values like positive, negative and neutral values manually for each emoji based on the viewers during the game.

 

Day 2 & 3: I continued my week by increasing few more tweets and coming up with few rules that need to be worked on as the ideas for each individual varies from person to person and it should be done while studying tweets carefully.

Day 4&5: I worked assigning sentiments for the tweets by calculating the sentiment based on the values that were added on the previous days. And the sentiments showed positive results as per the demand, I required for the work.

It is satisfaction of the this work that made me happy for the project I have been working on. And yes, the calculation can be made in the upcoming week with better results.

WEEK-10

  WEEK 10:

Day 1: I started the week by discussing with the mentor regarding the different ways the emojis are stored in byte level with the article written by David de Hilster. 

Day 2 & 3: I continued my week by changing the heir.kb file that was sent by the mentor for the accepting the emoji dict in the NLP engine, which allowed the emojis to be identified by the parsers.

Continued with adding attributes values to the emojis to make it generalized and attr=1 indicates that it is positive and attr=-1 indicates it as negative.

The above image shows the functions that were created with respective changes in their rules also.

Day 4: I spent the day by generalizing the functions too and with the suggestions from mentor, the functions will be generalized even further. And installed WSL and ECL for later work in the upcoming week.

WEEK 9

 WEEK 9:

Day 1: I started the week by creating the functions necessary for generalizing the rules that I created the previous week. I faced few issues while executing them as it took few trials and errors.

Day 2 & 3: I continued my week by solving my issues with the problems assigning the emojis to the knowledge base, and it was such a minor error that I did like in the initial week and yes, it is the hard way to learn the mistakes we do. The knowledge base looks as described below with 

• Player name 
• Emoji Type
• Type of Sentiment

The above image shows the functions that were created with respective changes in their rules also.

Day 4: I spent the day with the article written by the mentor "https://hpccsystems.com/resources/the-hidden-world-of-unicode-from-text-to-emojis-2/", which helped me understanding the representation of emojis in the Unicode UTF-8, as the emojis integration into the NLP++ based on the skin tone and how it was represented in modern world.   

WEEK 8

 WEEK 8:

Day 1: I started the week by creating rules to excise the unnecessary tweets from the text, which needed some work to be done for the day. 

Day 2 & 3: I continued my week by creating the rules for applying the sentiments for the tweets and it was trial and error basis. I tried to focus on creating them general for all the tweets, but was able to make them exact. This needed to refactored for the next steps.

                The above image shows the rule for the user sentiment, which is quite exact for now. 

            The above image shows the rule for the cricket sentiment, which is generated through the cricket terms that are stored in the dictionary created previously. 

            The above image shows the rule for the team support sentiment, where the sentiment values are assigned based on the support received by the team through the emojis. 

   The rules created are exact and needs to be refactored by creating dictionaries for the user sentiments and making a knowledge base for the emojis and the sentiment values separately which can be used in future for evaluation of the final sentiment.

WEEK 7

 WEEK 7:

Day 1&2: I started the week by making a list of sarcasm statements that are generally used for the tweets regarding the team. The list would need to be updated regularly as the views of the individuals vary from time to time, but we are focused at the particular point of time.

    The above dictionary image represents the emoji dictionary that was built 

    The above image shows the player sentiment that was mapped along with the emoji rather than the rule used in the previous week.

Day 3: On meeting with the mentor, during the start of the week and taking his inputs, I continued completing the emoji dictionary as the emoji functionality in NLP++ has been updated, which would be of great help for my continuation of the work. I continued my week by removing the unnecessary tweets by creating general rules for it.

WEEK 6

  WEEK 6:

Day 1&2: I started the week with creating the sentiment values to the tweets, by creating keywords in the dictionaries. The images represent the emotions attached and the corresponding keywords on the right.

      

Day 3: On meeting with the mentor and taking his inputs, the previous work was not the NLP++ was made for as I worked on algorithmic approach which is not the correct one, so I focused on the starter analyzer provided by the mentor and worked on understanding the tweets on how a human would think of them. Later, I started building on the analyzer by adding rules for extracting account tags and non- account hashtags, which are provided below in the screenshots.

    

Day 4: This day I dedicated myself on creating the emoji dictionary and rules for emoji's as the dictionary wasn't supported, but thanks to the mentor who would be providing the option for emoji dictionary which would be of great help. I even updated my player as specified by the mentor. The images below represent the emoji dictionary and rules for player sentiment along, with the player updated lists.

 

WEEK 5

 WEEK 5:

Day 1&2: This week, I started my actual work of the project after completing few weeks of understanding the concepts of NLP++. I started with converting the tweets obtained into txt files that can be used easily for the ongoing work. The figure below depicts the part of the tweets that I was working on.

Day 3: The day went through going through Pedro's work on the emotion analysis that he worked on and tried to understand his way of working with NLP++ and some areas of how he has been using the debugging. Later I worked on creating the dictionary on some of the cricketing terms and the teams that usually play in the Indian Premier League.

 The two images shows the dictionaries made on cricketing terms and the teams played in the league.

Day 4: This day I dedicated myself on creating the huge list of players played for the team of Royal Challengers of Bangalore(RCB). I found it difficult while creating this list where I was filling up the details one by one. Later, the idea of why don't I use NLP++ to create the text that is required for the dictionaries made my work easier, where I created another analyzers for them and completed my task that I intended to.

This week's work was satisfying and easier after learning those concepts over past few weeks which made the work simpler and faster.