A fast introduction to Structure queries

Welcome ! Let's learn how best to make queries in structure

Introduction

Structure relies on a large database of pre-processed scholarly documents. Each of these documents has gone through multiple stages of processing to extract its contents and how these elements were related. We turned the papers into rich structured data.

Through this process, we discovered not only how the ideas and concepts in a paper relate to each other internally, but also how they relate to other papers.This creates a large network of relations between ideas and concepts, from the most basic unit of meaning to high levels of abstraction

The key benefit of structure is to make it easy for you to query such a large and complex network of relations. Now that we understand what we're searching, let's deep dive into the search itself.

Querying

We've seen above that Structure converted unstructured data into structured data. Its basic unit is a relation.

Basic unit

A relation is a triple between two entities, with a relation type. This abstract schema is agnostic to the type of objects or their relationships. Subjects, objects and relations can be of any type.

A basic relation is already a query, and can directly query similar relations in the corpus.

Building structures from basic units

These relations can be concatenated into a larger structure, by using another relation as operator.

Formulating complex queries

Our goal is to give you the utmost flexibility for querying.

    You have three main options:
  1. - Natural language : for basic queries
  2. - Using SPARQL : to access with customizable code
  3. - Using structureQL : for an easier yet customizable interface
structure automatically detects the type of langage you are using, no need to specify.

Natural language

You can query structure exactly as you would with other academic databases. You'll type in the relevant concepts and we'll convert it to a structure of relation.

SPARQL (alpha version)

Using SPARQL requires some familiarity with and to go through the structure ontology. We use a basic ontology, but in the future we plan to make it customizable. The advantage is that you have full control.

structureQL (alpha version)

structureQL is a simplified version of SPARQL, which allows you to mix SPARQL elements with natural language - in effect combining the above two types of query.