Search

Protogrid provides a powerful search functionality in tableviews and selection/dropdown lists. It is useful to find a Card for which characteristic text is known (e.g. a product name mentioned in an advertising text). If you want to comfortably restrict the display of Cards based on certain field values (e.g. categories or status), our Filter Fields might be the better choice.

Single text values can be entered straightforward. If you would like to search for word combinations enclose your search with double quotes (as you would do in Google).

Searches for number values do not require any special handling.

Searching freely for date and time is not recommended because of ambiguities with time zones. Usage of our dedicated Date Filters might be the better choice.

When a Neural Search is activated and a configuration is set, you can search for cards based on meaning instead of exact words.

To set a configuration go to “Application Properties” > “Neuralsearchconfigurations”. Here you can create a new configuration and set its name, embeddingmodel, language and chunksize. The embeddingmodel how the meaning of each card is created and the following models are available:

• embeddinggemma:latest
• nomic-embed-text:latest
• qwen3-embedding:0.6b

The chunksize controls how large a card can be in characters, before it is split up into multiple chunks. It makes sense to set the chunksize depending on how large individual parts of information are. This means for longer coherent texts a larger chunksize is recommended. A good starting point is at 3000.

It is recommended to set multiple configurations and try them out. The first matching configurations for the users language will always be used. If no configuration for the current user's language exist the first one will be used. This means the order of the configurations matters.

The longer and more precise your query is, the more accurate your results will be. The results will always be ordered by relevance.

A neural search can be initiated by:

• By clicking the brain button next to the lens button for full text search
• By pressing shift enter
• By including the keyword “NEURAL” in the beginning of your search query

Notes: Neural search is currently in closed beta status. If you want to join, feel free to contact us at Protogrid Support. We are also happy to hear about feedback after testing this feature.