Overview of the Filter tool

The Filter tool is designed to refine large datasets, making it easier to focus on relevant documents. You can apply metadata filters for attributes like publication dates and repositories, context filters based on detailed thematic descriptions, and analysis filters derived from dataset concepts and topics. Additionally, it allows sorting by criteria such as relevance or full-text availability, ensuring efficient prioritization of results.

Read Transcript

To use the Filter tool, you just have to click on your dataset and then the Filter tool on the right hand side panel where all tools reside. Then the Filter panel will open up where you can add and adjust filters.

There are different types of filters you can apply: metadata filters, context filters and analysis filters.

In metadata filters you can choose the range of publication dates by moving the markers left or right or decide to either include or exclude certain repositories. To include the repository, you need to write down the name and click on it - it will appear underneath. To exclude a repository, you need to include it first and then click on it again, so it will change color. You can also Filter the list by the full-text availability.

Here, you can use context filters, which are natural language descriptions of what content you are looking for. If you click “select filters”, you can choose from already existing ones. If you want to create a new context filter, just click on “+ new filter” on the right side. It is simple to create a context filter. You need two things: a name for the filter so you can find it later, and a description in your own words. The filter will be automatically saved so you can reuse it later or on different datasets.

A context filter can also be used both for inclusion - meaning high matches with the text - or for exclusion - meaning setting the scores of the context similarity low.

When you have run the Analyze tool on a dataset, you can use the analysis filters. You can choose keywords, concepts and topics for inclusion and exclusion. These words and topics are generated from the actual content of your articles.

The concepts are one-word representations of an idea, rather than keywords. The machine will include or exclude articles that include the same concept - not the same exact word.

With keywords and concepts you can use Boolean logic. Next to the included concepts you can choose between operators “AND” and “OR” to specify how you want to filter the results.

Regarding the topics, these are again generated by your starting document collection. The tool has created groups of articles within your dataset, and named them with a string of ten concepts that needs to be read in unity. Keep in mind one article can be part of multiple topics as you use them to include or exclude from the list.

To apply these filters, just click on a drop down list and choose. To exclude, double click it like before.

Lastly, you can find Chat filters. Here you can filter your reading list based on references from answers provided by the Chat tool. To learn more about it, watch our Chat tool tutorial.