Classification

# **Classification**

Situation 1: Classify an incoming support ticket by urgency (critical, high, medium, low) and department (billing, technical, account management, general).  
———————————————————————————————————————  
Prompt 1: You are an expert support operations specialist responsible for triaging incoming tickets. 

First, extract the following from the incoming support ticket: customer\_name, message, date. The support ticket is provided below: 

\<support\_ticket\>{{support\_ticket}}\</support\_ticket\>

Then, categorize the ticket into the correct urgency group and department.   
\- For urgency groups, categorize the ticket into one of the following: critical, high, medium, low.  
Use these criteria for urgency:  
\- critical: system down, data loss, security breach — immediate business impact  
\- high: major feature broken, significant revenue impact, SLA at risk  
\- medium: partial functionality affected, workaround exists  
\- low: general questions, minor issues, cosmetic problems  
\- For department, categorize the ticket into one of the following: billing, technical, account management, general.

Return the result in JSON with exactly these keys: customer\_name, message, date, urgency, department.

If a field is not present in the ticket, return null for that key.  
———————————————————————————————————————  
Situation 2: Classify a product review as positive, negative, or neutral — and if negative, tag the specific complaint category (shipping, quality, customer service, pricing).  
———————————————————————————————————————  
Prompt 2: You are an expert support operations specialist responsible for dissecting product reviews.

First, extract the following from the product review: customer\_name, product\_name, message, date. The product review is provided below:

\<product\_review\>{{product\_review}}\</product\_review\>

Then, based on the extracted message, assign one of the following sentiment options to the product review: positive, neutral, negative.  
\- Positive: The customer clearly displays positive emotions (joy, happiness, gratitude, fulfillment, etc.) in the message  
\- Neutral: The customer neither praises nor complains — factual or indifferent tone with no strong emotion either way  
\- Negative: The user has complaints and displays negative emotions (frustration, confusion, disappointment, etc.)

If the product review is deemed negative, categorize the review to one of the following based on the complaint that the review displays: shipping, quality, customer\_service, pricing.

Return the result in JSON with these exact keys: customer\_name, product\_name, message, sentiment, date, complaint\_category.

If sentiment is positive or neutral, return null for complaint\_category. If a field is not present in the review, return null for that key.  
———————————————————————————————————————  
Situation 3: Classify a job applicant as strong fit, potential fit, or not a fit based on their cover letter and a provided list of role requirements.  
———————————————————————————————————————  
Prompt 3: You are an expert interviewer responsible for analyzing applicants and deciding whether or not they are a strong fit for corresponding jobs and positions.

First, review the applicant's materials in the job application below, paying attention to their cover letter, resume, and interview transcript. The job application and list of role requirements are provided below:

\<job\_app\>{{job\_app}}\</job\_app\>  
\<role\_requirements\>{{role\_requirements}}\</role\_requirements\>

Then, based on the resume, cover\_letter, and interview\_transcript, classify the application into one of the following: strong\_fit, potential\_fit, not\_fit.  
\- strong\_fit: The applicant has all the skillsets required, has a deep history (3+ years) of working in this field, and has demonstrated clear, structured communication and asked informed questions during the interview.  
\- potential\_fit: The applicant has some or most of the skillsets required, has some history (1-3 years) of working in this field, and has demonstrated their willingness to learn during the interview.  
\- not\_fit: The applicant has little to none of the skillsets required, has little to no history of working in this field, and has not demonstrated any quality soft skills during the interview.

Return the results in JSON with these exact keys: applicant\_name, date, fit\_level, reasoning (one sentence justifying the classification).

If a field is not present in the job application, return null for that key.