Search the clinics data guide with Knowledge

Upload a Markdown guide for clinics.csv to Knowledge Management and search for the information you need. This optional tutorial follows Create a dashboard from CSV data and is not part of the quick-start completion criteria.

Prerequisites

• A D.Hub account and a collection where you can add documents
• The my_first_project collection created in the Quick start
• clinics data guide

You can complete the tutorial with keyword search alone. To use semantic search as well, an embedding model must be available on the knowledge creation screen. Complete the Chat step only when an LLM model is available.

Synthetic training data

clinics.csv and its data guide are synthetic data for learning D.Hub features. Do not use them to verify real medical-facility information or specimen collection availability.

1. Download the sample

1. Download the clinics data guide.
2. Confirm that the guide describes the fields and 20 sample rows in clinics.csv.

The quick start uses the CSV to create a dataset and map dashboard. This tutorial uses the Markdown file describing the same data as a searchable document.

2. Create knowledge

1. Select AI → Knowledge Management in the sidebar.
2. Select Create knowledge at the top right of the list.
3. Enter clinics_guide as the name, then select my_first_project under Collection.
4. Under Search options, select Keyword search.
5. To retrieve similar expressions, also select Semantic search and choose an embedding model.
6. Select Create.

The Documents tab of the clinics_guide detail screen opens.

3. Upload the Markdown document

1. On the Documents tab, select Add document → File upload.
2. In the File upload step, add clinics-data-guide.md. Review the automatically populated document name and select Next.
3. In Chunking settings, keep the defaults. Confirm that the chunk preview contains the title and field descriptions, then select Next.
4. Under Review and submit, review the file, chunking settings, and search mode. If only keyword search is enabled, Keyword only appears.
5. Select Upload file.

The document list reopens after upload. Wait until clinics-data-guide changes to Complete. If it remains Processing for a long time or changes to Failed, review the error in the document row.

4. Review search results

1. Select the Search tab.
2. Enter each question below and run the search.
3. Confirm that clinics-data-guide.md appears in the results.
4. Open the source chunk and confirm that it matches the information in the table.

Search question	Expected content
Which field indicates whether specimen collection is available?	The sample_available field and the meaning of Y and N.
Which fields should I use to show the facility location on a map?	The latitude and longitude fields.
Which fields are used to filter by region?	The sido, sigungu, and dong fields.

If there are no results, first confirm that the document status is Complete. For keyword search, search again with an exact field name from the document, such as sample_available or latitude.

5. Review an answer in Chat (optional)

Complete this step only when an LLM model is available.

1. Select the Chat tab.
2. Enter one of the questions used in Search.
3. Select a citation number shown in the answer.
4. Confirm that the source is clinics-data-guide.md and that the cited text supports the answer.

If no chat model is available, the tutorial is complete after reviewing the results and source chunks for all three questions on the Search tab. Chat answers summarize the synthetic data guide and must not be used as real medical information.

Verify the result

• The my_first_project collection contains the clinics_guide knowledge.
• clinics-data-guide shows Complete on the Documents tab.
• The Search tab shows field descriptions and source text from clinics-data-guide.md.
• If a chat model is available, the answer citation shows the same source text.

To further adjust the search method and result scope, see Review search results and Chunking and search options.