Frequently Asked Questions
What does Hubscience do?
Hubscience helps you build up your knowledge graph over your topic while reading scientific publications.
What is the knowledge graph?
It is a bunch of connected information from which a researcher can derive new knowledge.
Does Hubscience creates the knowledge graph automatically?
No, HubScience uses a hybrid model, an expert needs to do the first steps manually (manual annotation), this will teach the annotation engine so the rest of the documents are annotated automatically.
What should I do as first steps?
If you start to work on a completely new topic, you have to create a new project. Then, you have to start to read and annotate the important terms manually: highlight the text and categorize them. These terms will be added into wordpacks and Hubscience will reuse reuse them from now on when analyzing other topics or documents.
If you cooperate with other colleagues you can share the documents and the manual annotation task with them to accelerate the training procedure.
I manually annotated a few article and automatically annotated others. Can I have an overview of the annotations?
On the documents page, you can find the number of annotations on every document card.
Besides, on the knowledge page, there is an elaborated statistics with filtering options.
I have a lot of annotations, still not know why it is a knowledge graph?
You need to do another step while annotating, to make relation between annotations. Two annotations can be connected if they belong to each other. In the end, these connections (edges) will turn annotations into a graph.
I see the graph, but why is it a knowledge graph?
Using the right labeling on the edges between the annotations makes the annotation graph a knowledge graph.
What is the good labeling of annotation connection?
It is not difficult, you only need your common sense. To understand this I give you an example here.
You probably heard about algopyrin, but maybe have not heard that its active agent is called metamizole. Let's suppose you read an article about one of its side effects: hypersensitivity.
You annotate the metamizole and the hypersensitivity and connect the two annotations as the following.
You see the given label is 'has side effect'. This way the two nodes and the label can be read out as a sentence.
Metamizole has side effect: hypersensitivity.
How will information fragments be chained to each other?
Imagine you have a series of information fragments like above and even they are all connected to each other like this.
How do annotations and their relations build a graph?
You can connect one annotation to several other annotations. They don't need to be close in the text, actually, you possibly find them in different publications.
If we just go on a bit further with our examples we might get this graph easily.
This is the knowledge graph, connected information.
How can I derive new information from the knowledge graph?
The recipe is:
Read out the graph starting from any point the graph and you can choose any arrow to continue reading.
All cases you will have competent, valuable sentences.
Some will be well known by you, maybe it was you who made the connections. Others come from different parts of the publications, made by another member of your team and form new, non-trivial information.