Traditional search is built around string matching. tagtog uses your annotated data to create a semantic layer allowing users to perform queries across their documents using concepts, labels and other metadata. The augmentation of the search functionality makes easier to discover patterns or find actionable insights. This is a major benefit when you have built a model that annotates text automatically and you want to grasp the intelligence of the data processed.
The search engine can be used through the user interface or the API. Learn how to build queries and make the most of the concept search.
You can find the search bar in the main toolbar from the web app. Here you type search queries.
You can access the advance search panel by clicking on the arrow on the right side of the search box. This panel uses the data from your project settings (dictionaries included) to automatically create search queries friendly.
Find which documents have been already marked as confirmed or which not.
Find which documents contain at least one annotation from the entity type selected from the dropdown menu.
Just start typing the first 3 characters of one of the names of the entity you are looking for. All the possible entities gathered across all your dictionaries will show up. Click on one of the items to display the search query on the normalization text box and, when you are ready, just click on the Search button to retrieve all the documents that contain at least one entity normalized to that name. In the case of the picture below, all documents containing entities with the name: minivan, people carrier, etc.
The search results are paginated. If your query retrieves a long list of results, you will find a
Load more link at the bottom of the list. Just click it to load more results.
Documents such as biomedical articles have usually associated an id (e.g. PubMed articles). Type the id in order to find matching documents in the pool. You can also use wildcard characters as in the example below.