Build Enterprise Search using Vertex AI Search and Text Summarization using Google Generative-AI Model for IT Incident of SMartApp. The figure and description below how it works.
- Retrive incident and detail data from database to create ndjson file.
- Ingest recently created json file to Google Cloud Storage.
- Import json file from Google Cloud Storage to Search DataStore increamentally on daily basis.
- On The Search Page ,try search something such as brand,model,incident type,subject, description on Search page to get the result.
- Additionally, you can use the Facet-Widget below the search text box to narrow down to get more accurate results.
- Try search something on Search page and click to view search results detail.
- Click Summarization on Search page
- Grab entire incidence detail to create incident content template to let Gen-AI generate incident summary.
- Store incident summarization to SmartApp database including summarization feedback (like/disklike) from user for 2 main purposes.
- Evaluate how well the model generate summarization.
- Collect incident summary to classify the incident knowledge base as a knowledge repository in the future.
- RunImportData.py : Run 3 main module files in IncidentSearch_DataProcessing sequcially as above process description in Incident Search (Vertex AI Search)
- IncidentSearch_DataProcessing : 3 Core module files to ingest json data from database to cloud starage and data store in Searh App accordingly.
- IncidentSummarization_API : Module files and template files to generate incident summarization as specified template description, It is service as FLASK API in order for SMartApp WebApplication to invoke this function .
- IncidentVectorSearch : Implement Vertex AI Vector Search to return the top 5 of the most relevant incident subjects corresponding to your keywords on the search page.