badrelfarri / build-ai-enriched-corporate-training-catalog

Udacity third project: Build AI Enriched Corporate Training Catalog

Building a cognitive search solution for searching in courses databases and articles within the organization.

Business Needs:

• Users can search the course data by searchable description.
• Users can search the course data for course rating and duration.
• Users can facet course data by source, skills, and role.
• Users can search the curated library for papers by searchable content (document cracking using OCR).
• Users can search the curated library for papers by authors' name, institution name, publication name, publisher, and publication date.

System Capabilities (Requirements):

• Users should be able to search all text fields for keywords.
• Users should be able to filter on course ratings, duration, title, type, source, level, technology, rating, instructor, or role.
• Users should be able to filter and facet on source, skills, or role.
• Users should be able to sort by relevant fields, such as title, type, technology, role.
• Users should be able to search and filter by keywords and phrases, authors, and institutions based on the contents of a PDF as inferred from an AI enrichment of the Document Extraction.
• Users can search the course data by keywords and phrases from descriptions and other fields.
• Users can filter course data results based on relevant criteria such as course ratings or duration.
• Users can see results from searches of course data for the course instructors for their own internal training.
• Users can facet course data by level, product, or role.
• Users can sort course data search results by relevant fields.
• Users can search the curated library for papers by keywords and phrases from the Journal name, authors, and their associated institutions.
• Users can search the curated library for papers by publication name, publisher, and publication date.
• Users can see results from searches of the curated library for papers with DOI (Digital Object Identifier) URL for the file.

Architecture of the solution

Both data sources are stored in a blob storage. The course data is stored as tabulated data while the pdf articles are stored in a container blob.