Models help you structure your work with data by providing several objects and functions. The key ones are Dataset and Record – a Dataset being a collection of Records. Additionally, there is a a Field object for describing the columns of a Dataset, a Query object for describing queries, and a Facet object for holding summary information about a Field (or multiple Fields).

All the models are Backbone models, that is they extend Backbone.Model. Note, however, that they do not ‘sync’ (load/save) like normal Backbone models.


A Dataset is the central object in Recline. Standard usage is:

var dataset = new recline.model.Dataset({
  // general metadata e.g. 
  id: ...
  title: ...
  // information about data source e.g.
  // backend string or object
  backend: a string identifying the backend we are using - see below

// initialize dataset with data from the backend.

// we will now have the following (and more) set up - see below for details 
dataset.fields // collection of Fields (columns) for this Dataset
dataset.records // collection of Records resulting from latest query
dataset.docCount // total number of Records in the last query

Key Attributes



queryObj is an object following the query specification below.

Record (aka Row)

A Record is a single entry or row in a dataset. A Record needs little more than what is provided by the standard Backbone Model object. In general, you will never create a Record directly – they will get created for you by Datasets from query results.

Field (aka Column)

A Field should have the following attributes as standard:

var field = new Field({
  // a unique identifer for this field- usually this should match the key in the records hash
  id: 'my-field-id'

  // (optional: defaults to id) the visible label used for this field
  label: 'My Field Name',

  // (optional: defaults to string) the type of the data in this field.
  // For list of type names see below
  type: 'string',

  // (optional - defaults to null) used to indicate how the data should be
  // formatted. See below.
  format: null,

  // (default: false) attribute indicating this field has no backend data but
  // is just derived from other fields (see below).
  is_derived: false


The type attribute is a string indicating the type of this field.

Types are based on the type set of json-schmea with a few minor additions and modifications (cf other type lists include those in Elasticsearch).

The type list is as follows (brackets indicate possible aliases for specific types - these types will be recognized and normalized to the default type name for that type):

NB: types are not validated so you can set the type to whatever value you like (it does not have to be in the above list). However, using types outside of the specified list may limit functionality.

Rendering, types and formats

One can customize the rendering of fields in the user interface and elsewhere by setting a renderer function on the field. You do this by setting a field attribute:

myfield.renderer = myRenderFunction;

Your renderer function should have the following signature:

function(value, field, record)

Where the arguments passed in are as follows:

Note that implementing functions can ignore arguments (e.g. function(value) would be a valid formatter function).

To guide the behaviour of renderers we have type and format information. Example types and formats are:

Default renderers are provided - see the source for details, but a few examples are:

Derived fields

Some fields may be ‘dervied’ from other fields. This allows you to define an entirely new value for data in this field. This provides support for a) ‘derived/computed’ fields: i.e. fields whose data are functions of the data in other fields b) transforming the value of this field prior to rendering.

To use derived fields set a deriver function on the Field. This function will be used to derive/compute the value of data in this field as a function of this field’s value (if any) and the current record. It’s signature and behaviour is the same as for renderer.


Query instances encapsulate a query to the backend (see query method on backend). Useful both for creating queries and for storing and manipulating query state - e.g. from a query editor).

Query Structure and format

Query structure should follow that of ElasticSearch query language.

NB: It is up to specific backends how to implement and support this query structure. Different backends might choose to implement things differently or not support certain features. Please check your backend for details.

Query object has the following key attributes:



Sort structure is inspired by but with some standardization.

Sort structure must be as follows:

"sort" : [
      { field: "post_date",  "order" : "desc"},
      { field: "user" },
      { "name" : "desc" },
      { "age" : "desc" },
      {"_score": null}

If order is omitted it is assumed to be “desc” except in the case of _score. _score is a special case which is used for match score if that is supported by the backend.


   q: 'quick brown fox',
   filters: [
     { term: { 'owner': 'jones' } }

Facet – Summary information (e.g. values and counts) about a field obtained by a 'faceting' or 'group by' method

Structure of a facet follows that of Facet results in ElasticSearch, see:

Specifically the object structure of a facet looks like (there is one addition compared to ElasticSearch: the “id” field which corresponds to the key used to specify this facet in the facet query):

  id: "id-of-facet",
  // type of this facet (terms, range, histogram etc)
  _type : "terms",
  // total number of tokens in the facet
  total: 5,
  // @property {number} number of records which have no value for the field
  missing : 0,
  // number of facet values not included in the returned facets
  other: 0,
  // term object ({term: , count: ...})
  terms: [ {
      "term" : "foo",
      "count" : 2
    }, {
      "term" : "bar",
      "count" : 2
    }, {
      "term" : "baz",
      "count" : 1