Frequently Asked Questions

What are the advantages of FriendlyData compared to similar products?

  • Accuracy. FriendlyData's Parser is 2 times more accurate than open source alternatives.
  • Simple onboarding. Doesn’t require training on the massive datasets. FriendlyData only need to know the schema of your data to get started.
  • Compatibility. Can connect to any relational database and customizable for the domain specific languages.
  • Fast integration. The integration is ~100 times faster than building an in-house solution based on NLP libraries.
  • Safety and data privacy Your data stays in your database, we do not move it or even index it.
  • Flexibility Easily customizable for any structured data source, it allows to introduce new synonyms and domain-specific rules and works with both: natural language and Google like search queries.

Is there any time limited trial version available to evaluate?

We are happy to build a custom demo or POC project for you.

Can you offer an on premise setup?

Both (cloud and on-prem) options are available. The main advantage of the cloud integration is faster support and ability to update solution on the fly. We have a very aggressive roadmap and the cloud solution would allow us to deliver updates to you without involving any resources from your side.

Tell me about the integration.

The integration process is really smooth and painless for you. You only need to provide FriendlyData read access to your database and after that we will make all required configuration and provide you with ready-to-go application. We need to know database schema to configure our system to parse requests in this specific case and to generate valid SQL queries.

What databases do you support?

We support a wide list relational DBMS (including PostgreSQL, MySQL, MSSQL, Amazon Redshift, etc) so we are pretty sure that if you have any relational DBMS then we support it.

Is my data safe?

Yes, your data is absolutely safe - we don’t require you to move your data anywhere and we even can work without any access to your data - we only need to know your database structure and columns names and types.

What languages do you support?

Our main focus is English2SQL but we are planning to extend our language coverage.

What’s the difference between FriendlyData API and open source NLP libraries? Can I build an in-house solution for databases using such libraries?

You can use a natural language processing library coupled with a rule driven post-processing module that analyzed the syntactical tree. But such approach has the following drawbacks: the input query has to be a grammatically correct sentence which was not always the case since the users often tend to simplify constructions and use symbols not typical for natural language, i.e. creating a mix of natural grammar and formal query language. Also, NLP parsers are not 100% accurate and the accuracy drops dramatically in longer sentences, e.g. when the query contains a chain of 4 or more conditions. FriendlyData’s query parsing module based on the proprietary formal grammar based technology. The technology features the following advantages that improve the technology learning curve and customer on-boarding time:

  • FriendlyData parses both natural language and any structured or semi-structured types of queries
  • The technology is based on a simple plain text grammar definition format that can be easily edited and compiled on the fly.
  • The grammar format allows for custom solutions that require introduction of new synonyms, data types and operations.
  • The parsing module outputs a tree specifically modelled for query analysis purposes which provides full control over the structure and makes the post processing and query object translation easy and transparent.

Learn more in our Blog Post

NoSQL?

We don’t support any NoSQL solutions right now, but we have very aggressive roadmap which includes implementing FriendlyData for NoSQL databases. You can subscribe to our newsletter not to miss any significant updates.

What about accuracy?

It depends on your particular case, but as we found out from our customers experience, the accuracy is about 90% in average. The accuracy approaches 100% for queries written in grammatically correct English.

Does FriendlyData ‘translate’ into SQL only or are there other options available?

FriendlyData translates into different SQL dialects but we can also respond with structured object which you can use according to your needs or we even can translate into your own DSL.

How sophisticated FriendlyData can get for a non-technical user to handle?

There is no limitation on the number of conditions in the query, so you can combine them as you want and make as complicated queries as you can imagine.

Can you handle question that are referenced to data in more than one table?

We can handle has-many and many-to-many (through one join table) cases. In such cases, we support different data types, aggregation functions, ‘like’ search and etc. In the nearest few months, we are going to rapidly improve our SQL generation module to support much more complex database structures.