Querying: More Like This


  • more_like_this returns a list of documents that are related to a given document.
  • This feature can be used, for example, to show a list of related articles at the bottom of the currently-read article page, as done in many news sites.
  • To accomplish this, RavenDB uses the Lucene contrib project MoreLikeThis feature.

  • In this page:


Setup

To be able to work, more_like_this requires access to the index text.
The queried index needs, therefore, to store the fields or the term vectors for these fields.

from ravendb import AbstractIndexCreationTask, MoreLikeThisOptions
from ravendb.documents.indexes.definitions import FieldStorage
from ravendb.documents.queries.more_like_this import MoreLikeThisStopWords

from examples_base import ExampleBase


class Article:
    def __init__(self, Id: str = None, name: str = None, article_type: str = None):
        self.Id = Id
        self.name = name
        self.article_type = article_type


class Articles_ByArticleBody(AbstractIndexCreationTask):
    def __init__(self):
        super().__init__()
        self.map = "from doc in docs.Articles select { doc.article_body }"
        self._store("article_body", FieldStorage.YES)
        self._analyze("article_body", "StandardAnalyzer")

Basic Usage

Many more_like_this options are set by default.
The simplest mode will satisfy most usage scenarios.

articles = list(
    session.query_index_type(Articles_ByArticleBody, Article).more_like_this(
        lambda builder: builder.using_document(lambda x: x.where_equals("id()", "articles/1"))
    )
)
from index 'Articles/ByArticleBody' 
where morelikethis(id() = 'articles/1')

more_like_this will use all the fields defined in an index.
To use only specific fields, pass these fields in the MoreLikeThisOptions fields property.

options = MoreLikeThisOptions(fields=["article_body"])
articles = list(
    session.query_index_type(Articles_ByArticleBody, Article).more_like_this(
        lambda builder: builder.using_document(
            lambda x: x.where_equals("id()", "articles/1")
        ).with_options(options)
    )
)
from index 'Articles/ByArticleBody' 
where morelikethis(id() = 'articles/1', '{ "Fields" : [ "ArticleBody" ] }')

Options

Default parameters can be changed by manipulating MoreLikeThisOptions properties and passing them to more_like_this.

Option Type Description
minimum_term_frequency int Ignores terms with less than this frequency in the source doc
maximum_query_terms int Returns a query with no more than this many terms
maximum_number_of_tokens_parsed int The maximum number of tokens to parse in each example doc field that is not stored with TermVector support
minimum_word_length int Ignores words less than this length or, if 0, then this has no effect
maximum_word_length int Ignores words greater than this length or if 0 then this has no effect
minimum_document_frequency int Ignores words which do not occur in at least this many documents
maximum_document_frequency int Ignores words which occur in more than this many documents
maximum_document_frequency_percentage int Ignores words which occur in more than this percentage of documents
boost bool Boost terms in query based on score
boost_factor float Boost factor when boosting based on score
stop_words_document_id str Document ID containing custom stop words
fields List[str] Fields to compare

Stop Words

Some Lucene analyzers have a built-in list of common English words that are usually not useful for searching, like "a", "as", "the", etc.
These words, called stop words, are considered uninteresting and are ignored.
If a used analyzer does not support stop words, or you need to overload these terms, you can specify your own set of stop words.
A document with a list of stop words can be stored in RavenDB by storing the MoreLikeThisStopWords document:

stop_words = MoreLikeThisStopWords(stop_words=["I", "A", "Be"])
session.store(stop_words, "Config/Stopwords")

The document ID will then be set in the MoreLikeThisOptions.

Remarks

Please note that default values for settings, like MinimumDocumentFrequency, MinimumTermFrequency, and MinimumWordLength, may result in filtering out related articles, especially with a small data set (e.g. during development).