There is a big difference between” Concepts” in Stage 1 and “Topics” in Stage 2. 

A concept is one word representing an idea that can be found in the papers you have loaded in the Focus tool. The machine has a conceptual understanding of the word and its contextual synonyms. Adding concepts for inclusion/exclusion will not only include/exclude papers with that exact keyword in it, but also papers using contextually similar words. 
A topic is a string of words representing a group of documents. The machine is asked to group the papers you loaded into the tool - where one paper can be part of multiple groups. When the machine is asked to describe each group, it does so with a string of 10 words that have to be read together, and can not be altered. A topic is therefore associated with a specific and numbered group of papers, and when a topic is included/excluded, that exact group of papers is the one included/excluded. 

.

In Stage  1, when including/excluding based on concepts, focus mostly on what you already know you want to exclude.

Having worked either with the Explore tool already, or with papers in your own remote folder, you likely have some ideas of what kind of noise your set of papers contain. Use this knowledge to add relevant exclusion criteria.

.

The Concepts in Stage 1 are not keywords.

You are not just telling the tool to include papers with this keyword, but to exclude papers that talk about the concept - the tool has a broader understanding of the concept than what it shows you in the one-word representation. 

.

Include always override Exclude, so be mindful of not adding broad concepts for inclusion in Stage 1.

Spend your time mostly on the known Exclusion criteria and only add very specific Inclusion criteria. 

.

Use the Focus tool iteratively in both Stage 1 and 2.

We recommend adding some criteria for exclusion, then reviewing the results (both the Excluded and Included papers) to see if there’s anything standing out to you - then go back to the concept/topic list and add more details.  

.

The more papers you manage to exclude in Stage 1, the more precise your Current topics will be in Stage 2.

Especially if you have a longer list of papers (i.e. 1000+ papers) with a lot of noise, spending some extra time in Stage 1 to narrow it down will allow the topic model in Stage 2 to be more helpful.
You may also choose to finalize Stage 1 to the best of your ability, move to Stage 2 and review the “Current” topics briefly before going back to Stage 1 to add more concepts for exclusion based on what you learned. 

.

In Stage 2, take time to review all topics for inclusion/exclusion.

The more inclusion/exclusion criteria you add, the more precise results you will have. However, here too, be mindful of not adding too broad Topics for Inclusion, as inclusion criteria will override exclusion. 

.

There are synergies between Explore and Focus

Go here to learn about them.

Tags:
Created by Iris Admin on 2020/11/30 16:17
© 2015-2020 Iris.ai AS. All rights reserved.
v1.0