rdf-dataset-textsearch v0.1.0

rdf-dataset-textsearch

This package provides text search for RDFJS DatasetCore compatible datasets.

Usage

The package exports a function to create a new dataset or wrap an existing one:

const textSearchDataset = require('rdf-dataset-textsearch')

const dataset = textSearchDataset(options)

The function supports the following options:

• dataset: The dataset that should be wrapped. If this option is not provided the factory will be used.
• factory: A RDFJS dataset factory that should be used to create the wrapped dataset. If the dataset option is given this option will be ignored. By default rdf-ext is used.
• properties: An iterable object that contains the definition for the search index. Each item must be an object with a term property. The value of the term property must be a RDFJS Named Node with the IRI of the predicates which should be indexed. Optionally a weight property can be used to weight the property for the search. By default the weight value is 1.0. In case multiple indexes are used '' (empty string) is used as indexId.
• indexes: An object that contains key-value pairs for multiple indexes. The key is used as indexId. The value must have the same structure as described in the properties option.

The function returns a dataset that implements the RDFJS DatasetCore interface. All methods of the spec can be used as described in the spec. Additionally there is a .search method to make a text search on the indexed properties:

const textSearchDataset = require('rdf-dataset-textsearch')
const namespace = require('@rdfjs/namespace')

const ns = {
  rdfs: namespace('http://www.w3.org/2000/01/rdf-schema#')
} 

const dataset = textSearchDataset({
  properties: [{
    term: ns.rdfs.label,
    weight: 1.0
  }, {
    term: ns.rdfs.comment,
    weight: 0.5
  }]
})

// fill the dataset with quads

const results = dataset.search('this is my query string')

results.forEach((result, index) => {
  console.log(`subject: ${result.value} (${results.scores[index]})`)
})

The .search method expects a query string as the first argument.

The method returns the subject of the matches as an array of RDFJS Terms. The best matches appear first. With the .scores property in the result, it's possible to get the score of the results. The value of .scores is an array with the same number of elements as the result itself. The score for a specific result can be found with the same index as the result (see the example above).

Optionally a second string argument for the indexId can be given. See the indexes options for more details how to create multiple indexes. By default the '' (empty string) index is used.