Data Discovery - Files Table - ITAG AI Search

A video of this support file is available here.

An Information Tag (ITAG) AI-based search feature has been added to the Files Table. This returns files with matching ITAGs across all projects to the Files Table in the Data Discovery page. The Map, Project Table and Files Table are linked, so the filtering in the Map or Projects table will update the Files Table, thus enabling the user to apply filtering to find specific groups of files in the search results.

The Information Tag Search Box is at the top centre of the Files Table. Users can type in a question or ITAG to help them find types of data. The Search system uses a large language model (LLM) to select the most appropriate ITAGs for whatever the question is. All files of the various ITAGs are then returned to the Files Table, and the user can then filter the data based on spatial selection in the Map, or by filtering columns in the Project and Files Tables. 

This search feature is not visible or available to Public Users and is currently marked as a Preview feature. The search uses an AI Agent and a large language model to provide answers; it may therefore return incorrect ITAGs.

Users start searching by typing a query in natural language into the Information Tag Search Box. The search button will not become live until the user has entered a query into the Information Tag Search Box. The query can also just be the name of an ITAG or CTAG, or just a grouping like “all seismic files”.



Once the Query has been entered, press the search button, and Files will start to load to the Files Table.

Please note that loading “all files” into the table will load > 1.3 million rows into the file table for further filtering and discovery. A user's local IT environment and hardware may impact performance where large proportions of the data collection are being accessed.

The other buttons in the Search interface will appear once the results have been returned

• “Search” performs the search based on the entered user query
• “ITAGs found” displays ITAGs that match the search query
• “Show on Map” displays results on the map
• “Clear” clears the search query from the text input

The “ITAGs found” button displays the Information Tags matched by the search query and a Confidence Score. The table is sorted by decreasing confidence score. A higher confidence score indicates that the LLM thinks the ITAG is a better match to the answer to the User query. Only ITAGs returned with a confidence score greater than 0.7 are loaded into the Files Table. The following screenshots show some examples of queries and their returned ITAGs. 

Depending on the query, many files can be loaded to the Files Table, for example, for one query string, the search resulted in 470,451 rows being added to the Files Table.

After the search has completed, clicking “Show on Map” displays the location of the files from the query on the map, and the button name updates to “Hide on Map”, which would remove them from the map display. 

Displaying files on the map plots a data density layer when the number of results exceeds 50,000. The data density layer shows the number of files per spatial hexagon, the colour coding of the hexagons corresponds to quartiles in the file density distribution, and the number can be read on the map legend. 

Other map sources can be displayed alongside the file search results. The image below shows a zoomed-in file density plot with the Carbon Storage Licence GIS Layer overlain.

Any outline or free-drawn polygon can be selected on the map. For example, selecting a Carbon Storage Licence will filter the Project IDs and Files Tables based on the outline of the polygon. The map will only update when the “Update Map” button is pressed.

The map will only update the search results layer when the “Update Map” button is shown and the user clicks it.

Now that the map item is selected, the “Update Map” button can be pressed to show the filtered files on the map. The filtering reduced the number of results in the Files Table to less than 50,000 results, so the file geometries are shown on the map instead of a file density plot.

Selecting a file on the map updates the map to show the selected file. For this SEG-Y file the pop-up has buttons to preview the SEG-Y data or SEG-Y header.

Clicking on the “SEG-Y data” button on the pop-up will open the seismic preview window and load a subset of the data to preview the SEG-Y data.

Example of seismic preview window opened from the map file pop-up

Further filtering can be applied in both the Project and Files Tables from any of the column filters; for example, filtering down by file type of “LAS” and MTAG of “sonic”, will reduce the number of files shown in the Files Table. Subsequent clicking of the “Update Map” button will show the files as points on the map. These files can be selected on the map by clicking on them, which will open the file information pop-up on the map, simultaneously the files table will update and scroll to show the file as the first row in the Files table.

A spatial query can be drawn as a freehand polygon on the map using the polygon map tool. The Project ID and Files Tables will then be filtered based on the user polygon. In this example, the number of rows has reduced to 983.

Clicking “Update Map” will show the geometries on the Map of the files currently shown in the Files Table.