Extracting Perspectives from Review Findings

Overview

This is a sample that takes multiple Excel files containing review findings accumulated in past projects as input,
analyzes the content of the review findings, and generates a list of review perspectives.

It extracts and organizes the points to be checked from individual review findings, eliminates duplicates, and lists them as a review perspective list that is easy to reuse.

• In this sample, processes such as file enumeration, finding extraction, perspective generation, and perspective integration are broken down and executed as a workflow.
  This allows you to experience the process of processing a large number of review findings step by step by combining tool integration and an AI agent.
• The aim is to experience the process of using AI to reuse knowledge and reduce reliance on individuals in development by standardizing review perspectives, which tend to be subject to individual expertise and tacit knowledge.
  

What this sample can do

• Analyze multiple Excel files in a specified folder
• Extract review points from each review comment using an AI agent
• Consolidate review points spanning multiple files and comments, and eliminate duplication
• List the extracted review points and organize them in a reusable format
• Check trace information to see which review points were generated from which review comments
• Combine and execute multiple processes, such as tool integration and AI agent calls, using a workflow

Target Users

This sample is primarily intended for the following users:

• Those who want to improve review points by utilizing past review comments
• Those who want to improve review quality by standardizing review points and reducing reliance on individual expertise
• Those who want to experience repeated execution using a workflow
  

Contents Included

This sample includes the following data as an example:

Type	Content
Workflow Definition	Review Perspective Extraction.yaml
Agent Definition	Perspective Extraction from Review Findings.yaml
Input: Review Findings List File	inputData/Excel files of Review Findings List
Documentation	This README

Using Sample Data

Prerequisites

• The Excel file containing the review findings to be analyzed is located in the designated folder.
• The content of the review finding contains the information necessary for extracting review perspectives.

Input/Output

• Input: Multiple Excel review finding list files in the inputData folder.
• Output: Review Perspective List, Review Findings and Review Perspective Trace Information

Procedure

1. Click "Review Perspective Extraction" on the Agentiqs workflow screen.

2. On the Review Perspective Extraction screen, click "Execute".

3. A dialog box will appear. Enter the "Review Findings List File Path" and click "Execute".

4. After the workflow is complete, the "reviewPointList.md" file will be output to the outputData folder.

Target Operation

• Workflow: Review Point Extraction
• Agent: Point Extraction from Review Findings

Limitations

• The generated review points depend on the content and granularity of the input review finding list.
• If the review finding description is ambiguous or lacks contextual information, appropriate review points may not be extracted.

Reference: Workflow Execution Flow

This sample demonstrates how to break down the process into tasks using a workflow and execute a series of processes by combining tools and AI.

1. Enumerate the Excel review finding list files in the specified folder.
2. Read each review finding list file and retrieve the review finding.
3. Input each review finding into the review point creation agent and extract the review points.
4. After all extractions are complete, perform integration and deduplicate removal using LLM.
5. Create a point list and trace information from the review points after deduplicate removal.