Challenge Submission Qhialdy Syira | qdiw

[qdiwasjati]

Hi Sector / Supertype team

Thank you for providing this challenge. it really help me to learn to know the components for a search engine AI tool which is very fascinating. i really am honored to join the 5th-6th feb session

Im still learning to code and swallowing sector's "frog" which is very loaded with information that im unfamiliar with this, so Im re-learning this with the help of claude sonnet 3.5 since it helps me understand code and concept that im unfamiliar with, hence i felt more comfortable stating its a collaborative effort.

Challenge 1, i asked claude sonnet for help what i understand was that we aim to ensure the user input goes through the fuzzywuzzy library so that the output of the user input (if mistyped) will still match the most similar subsector against the subsector slug list

steps:

• we update the def match_input_to_valid_subsector function as below:
• we create a loop to check each the similarities score of all the subsector slug against the user input
• for score of similarities above 80, the loop will include it into the good approximation list

Challenge 2, i asked claude sonnet again for help this was a bit tricky because we ended up updating also the def chat function so that it would use the fuzzy library if the input was not a ticker also added a flexible option if the request is a sentence.

steps:

• for the llm to use this tool we integrate also a loop inside the def chat function to use it when the user input is not a stock ticker exe: BMRI, BBCA
• if not then the def match_input_to_valid_subsector will be used to ensure we can provide the data best match with the user intention
• else the llm will execute based on the user input and add sectors data as additional insight

thank you soo much for the opportunity of this challenge. i really learned a lot and keen to learn more. thank you

Best,

Qhialdy

email: [email protected]
Google Collab: https://colab.research.google.com/drive/1SGGbHqr9HyUBpTA9_e5L52ZM9WYvc9JY

[onlyphantom]

hey @qdiwasjati ! I'll need the Google Colab docs to be open so I can view / run the code for grading.

I understand how difficult the challenge might seem, esp. for somebody relatively new in the field.

I truly appreciate the effort, and thank you for persisting till the end :)

[qdiwasjati]

Hi Sam,

thank you for the reminder, i have allowed the docs to be accessible, thank you for sparing your time for this!

https://colab.research.google.com/drive/1SGGbHqr9HyUBpTA9_e5L52ZM9WYvc9JY?usp=sharing

Best,
Qhialdy

[onlyphantom]

It looks like your search engine does reasonably well on ticker look-ups (INDF, TLKM, BMRI etc) but not exactly when you're asking for information on sectors. This could be seen from the output of your search engine when a sector enquiry is made:

→: beverages
🤖: A refreshing topic!
...

Some of the major players in the beverages sector include:

• PT Indofood Sukses Makmur (INDF.JK)
• PT Coca-Cola Indofria (INCOP.JK)
• PT Nestle Indonesia (NSTL.JK)
• PT Danone Indonesia (DANO.JK)
• PT Friskom Jaya (FRSK.JK)

Overall, for the work you put in, I feel inclined to mark this as a Pass, and issue a certificate here, but I feel that you can look at improving it! Thanks again!