IBM Deep Search

Version 7

This guide describes a quick and easy way to get up and running with IBM Deep Search.

Note

IBM, the IBM logo, and ibm.com® are trademarks or registered trademarks of IBM Corporation, registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at “https://www.ibm.com/legal/copytrade[Copyright and trademark information]”.

Product Overview

IBM Deep Search is a commercial Software-as-a-Service system that uses state-of-the-art AI techniques to index and explore collections of scientific and technical documents. Indexed data sources can include proprietary, internal PDF documents and databases as well as publicly accessible PDF documents and databases. Unstructured and structured data sources may be indexed, including embedded images, charts and tables. IBM Deep Search indexes these data sources and builds a searchable, browseable base of interconnected concepts, entities, and relations, derived from the data sources. Now you can spend minutes, instead of days and weeks, discovering new relationships, ideas and hypotheses in the easily searchable base provided by IBM Deep Search.

Overview and Prerequisites

The purpose of this Quick Start Guide is to teach you how to accomplish two basic tasks using IBM Deep Search:

  1. Executing searches on the millions of documents built into IBM Deep Search.

  2. Uploading your own PDF documents to be indexed by IBM Deep Search.

Note

An account on a SaaS instance of IBM Deep Search is required to complete this Quick Start

Time to complete this Quick Start: 10 minutes.

Step 1: Log into the IBM Deep Search User Interface

Point your Chrome or Firefox web browser to the login screen for IBM Deep Search, and complete the login using your valid IBMid.

Step 2: Go to the Library view of IBM Deep Search

Click on the menu link Library located at the top of the page, in the black menubar.

The Library view gives you access to the search capability, as well as access to all of the documents available in IBM Deep Search.

Note

Documents in IBM Deep Search are organized into sets called collections. Public collections contain publicly available scientific and technical documents that have already been indexed by IBM Deep Search. Private collections are collections that have been created by you, with your PDF documents.

On the Library view page you will see a Search box, and underneath the search box, three tiles:

  • Public collections: Click on this tile will expand it and present a list of all of the public data sources that have been indexed by IBM Deep Search and are available to you for searching.

  • Saved searches: When you execute a search, it can be saved for future use, and will be available under this tile.

  • Private collections: This tile will display all of the private collections of documents that you have created and saved.

Step 3: Execute a search using the Public collections

  1. Click on the tile for Public collections. This will bring it to the center of the page and open the listing of data sources available.

  2. Type some search terms into the search box. As you do so, you will notice the listing of public data sources will update to provide you with a count of matching documents from each data source.

Tip

As an example, enter the search terms carbohydrate metabolism liver

Step 4: Open the list of documents retrieved during search

A summary of search results will appear in on the Library view web page, under the Search bar.

The total number of documents identified will be displayed at the top of the Public collections list, labeled as Documents

As you scroll down the list of data sources, you will see the number of documents that were retrieved for each data source.

To browse through the documents retrieved from a particular source, click on the arrow to the far right of each source listing.

Tip

Example: If you entered the search terms carbohydrate metabolism liver, you should see a search result summary of approximately 15K documents from the source arXiv full documents.

Step 5: Browse the list of documents retrieved during search

Clicking on the arrow next to each source will bring you to the main search window of IBM Deep Search. This is where you can fine-tune searches as well as view the results.

The main search window is divided into three columns:

  1. Left-hand column: At the top is an arrow to take you back to the listing of public collections. The rest of the column contains controls for you to execute and fine tune your searches.

  2. Middle column: This column contains a list of the documents retrieved. Clicking on any of the links in this column populates the right-hand column with the specific information for that document.

  3. Right-hand column: Presents the content and metadata for each document when it is selected from the middle column. When viewing a document, there are three tabs that appear at the top of the right column:

    1. Description: Summary of the document, including the abstract and metadata.

    2. Document: The document content, parsed by IBM Deep Search.

    3. JSON: The JSON object created by IBM Deep Search after parsing the document.

Tip

The main search window provides many different features for modifying your searches and also for navigating the results.

Step 6: Uploading files to IBM Deep Search: Gather your documents

IBM Deep Search only accepts PDF documents for indexing. You will be uploading three PDF documents into the IBM Deep Search system so they may be parsed and indexed. You may use any three scientific or technical documents of your own choosing.

If you wish you may instead use the sample documents provided here.

Important

All documents uploaded to IBM Deep Search must be PDF documents!

Step 7: Create a new project.

IBM Deep Search collections are grouped together into projects. Projects represent the collections and other artifacts needed for your specific tasks or jobs.

To create a new project:

  1. Click on the menu bar link Community Project (upper left hand side).

  2. Click the button next to the text box labeled Create.

  3. In the text box labeled Enter a name, type a name for your new project (e.g., DS demo August 2023).

  4. Click the button under the text box labeled Create.

  5. The screen will then refresh with the Library view of your new project.

Step 8: Upload files into a new private collection

Now we can upload our PDF files into IBM Deep Search.

  1. From the Libary view of your new project, click on the tile Upload Files

  2. On the next screen make sure the radio button New collection is selected.

  3. In the text box labeled Enter a collection name, enter a name for your new collection (e.g., DS collection demo Aug 2023).

  4. Click the button labeled Upload.

  5. On the next screen you may drag and drop files to upload them, or click on the upload tile to bring up a file selection dialog.

  6. After a few moments of processing the screen will refresh with all green checkmarks that your documents have been "Successfully uploaded and processed".

  7. Click the Continue button.

  8. You will be taken to the Library search view of your new collection.

Step 9: Search through your new collection

Finally, do a simple search through your new collection.

Note

This Quick Start only involves a few documents. IBM Deep Search is capable of scaling to upload and index millions of documents! Search results are much more compelling when executed against millions of documents versus three documents.

  1. On the left hand side there is a text entry field labeled Search for text

  2. Click your mouse cursor in the space right under the Search for text label.

  3. Enter search terms related to the documents you just uploaded.

  4. Search results are presented as they were in Step 5 of this Quick Start.