Notes 3-18-25

https://uconn-sa.blogspot.com/ 

We were able to launch an app engine program from our compute engine instance. 

 I'd like to get all working code on the compute engine WM.

 john_iacovacci1@sentiment-prod:/home/sentapp$ ls appmenu dailyprices imdata

 john_iacovacci1@sentiment-prod:/home/sentapp$ 

We have created a directory for the application on /home/sentapp and use subdirectories for individual code.

 Nazmus had a github for Company info which we pulled into subdirectory and attached the app engine flask code by deploying to the sentiment analysis project app engine link.

 Problem is in order to run all the code from one app engine project linked all code needs to be merged together. 

 We need to understand how to use/grant access to the compute engine vm and datastore/firestore tables from sentiment test project.

Eventually both the SSGA code and the Webster code need to be merged but form development they can stay separated. APP Engine code for Company Info CompanyInfo and Company_info 

We created a new CompanyInfo table to be able to retrieve companies by ticker.

 Name/ID name=BAC Company_Name=Bank of America Industry=BANKS Ticker=BAC 

 All 30 companies on the list need to be added to this table. 

Topics For SSGA we are using the TECHNOLOGY sector or INDUSTRY (same field) from the 

Company info table to search for Keywords in the topics table to score Earnings calls.

Keyword Category Sector Client AI Agent AI TECHNOLOGY SSGA LLM AI TECHNOLOGY SSGA Model Training AI TECHNOLOGY SSGA

 For Webster we just have these keywords for now but we are only searching news.

We may expand to conference calls.

 Keyword Category Sector Client Customer service Customer Service BANKS WEBSTER Bank fees fees BANKS WEBSTER The Topics tables can be split or joined into one. 

We can keep it simple for now until we need one app engine. 

 Also, if we may be able to separate into two projects as long as Datastore and compute engine can be shared between projects. Daily prices All 30 companies from Company Info table need to be collecting prices on a daily basis into one single data store table in the Sentiment Analysis project.

 This should be running on a cron schedule In python code need to set project id Need project name set (your project PROJECT_ID = 'sentiment-analysis-379200’ And datastore kind client = datastore.Client() kind = 'dailyprices’ Eventually we will back fill prices. 

 Sentiment scoring SSGA. We need to follow similar model to last year with slight adjustments. Sentiment Detail Store details after scoring earnings call 

We use autokey Database ID (default) Namespace [default] Kind Sentiment_Details Key Sentiment_Details id:4503627796316160 Key literal Key(Sentiment_Details, 4503627796316160) URL-safe key ahtwfnNlbnRpbWVudC1hbmFseXNpcy0zNzkyMDByHgsSEVNlbnRpbWVudF9EZXRhaWxzGICAgPjogIAIDA 

 Call date Category Document Type Keyword Paragraph Period e.g. Q12025 

Paragraph Period Score Magnitude ' Score each paragraph that keyword appears in and store that paragraph in the scored record.

 After the score summarize in table for graphs Scoring summary Total each category that was scored Call date Document Type Period Ticker Category Score magnitude The Graphs need to use this table Need 8 quarters scored per company Graphs should be able to graph single company for 8 periods. Adjunct Professor John Iacovacci University of Connecticut, Stamford John.Iacovacci@uconn.edu