Embedding

Collaborative filtering as a ranker. Recommend is compatible with the library Implicit.

Parameters

• key ('str')

  Field identifier of each document.

• normalize ('bool') – defaults to True

  If set to True, the similarity measure is cosine similarity, if set to False, similarity measure is dot product.

• k ('typing.Optional[int]') – defaults to None

• batch_size ('int') – defaults to 1024

Examples

>>> from pprint import pprint as print
>>> from cherche import rank
>>> from sentence_transformers import SentenceTransformer

>>> documents = [
...    {"id": "a", "title": "Paris"},
...    {"id": "b", "title": "Madrid"},
...    {"id": "c", "title": "Montreal"},
... ]

>>> encoder = SentenceTransformer("sentence-transformers/all-mpnet-base-v2")
>>> embeddings_documents = encoder.encode([
...    document["title"] for document in documents
... ])

>>> recommend = rank.Embedding(
...    key="id",
... )

>>> recommend.add(
...    documents=documents,
...    embeddings_documents=embeddings_documents,
... )
Embedding ranker
    key      : id
    documents: 3
    normalize: True

>>> match = recommend(
...     q=encoder.encode("Paris"),
...     documents=documents,
...     k=2
... )

>>> print(match)
[{'id': 'a', 'similarity': 1.0, 'title': 'Paris'},
 {'id': 'c', 'similarity': 0.57165134, 'title': 'Montreal'}]

>>> queries = [
...    "Paris",
...    "Madrid",
...    "Montreal"
... ]

>>> match = recommend(
...     q=encoder.encode(queries),
...     documents=[documents] * 3,
...     k=2
... )

>>> print(match)
[[{'id': 'a', 'similarity': 1.0, 'title': 'Paris'},
  {'id': 'c', 'similarity': 0.57165134, 'title': 'Montreal'}],
 [{'id': 'b', 'similarity': 1.0, 'title': 'Madrid'},
  {'id': 'a', 'similarity': 0.49815434, 'title': 'Paris'}],
 [{'id': 'c', 'similarity': 0.9999999, 'title': 'Montreal'},
  {'id': 'a', 'similarity': 0.5716514, 'title': 'Paris'}]]

Methods

call

Retrieve documents from user id.

Parameters

• q ('np.ndarray')
• documents ('typing.Union[typing.List[typing.List[typing.Dict[str, str]]], typing.List[typing.Dict[str, str]]]')
• k ('typing.Optional[int]') – defaults to None
• batch_size ('typing.Optional[int]') – defaults to None
• kwargs

add

Add embeddings both documents and users.

Parameters

• documents ('list')
• embeddings_documents ('typing.List[np.ndarray]')
• kwargs

encode_rank

Encode documents and rank them according to the query.

Parameters

• embeddings_queries (numpy.ndarray)
• documents (List[List[Dict[str, str]]])
• k (int)
• batch_size (Optional[int]) – defaults to None

rank

Rank inputs documents ordered by relevance among the top k.

Parameters

• embeddings_documents (Dict[str, numpy.ndarray])
• embeddings_queries (numpy.ndarray)
• documents (List[List[Dict[str, str]]])
• k (int)
• batch_size (Optional[int]) – defaults to None