LycorisR is a lightweight recommendation algorithm framework based on LycorisNet.

Features

• Sequence-based recommendation.
• Discover correlations between items.
• Elegant and minimalist.

Installation

The project is based on LycorisNet, and the installation of LycorisNet can be found here.

pip install LycorisR

Documents

The APIs provided by Recommender (from LycorisR import Recommender):

Function	Description	Inputs	Returns
Recommender(config)	Constructor.	config: The configuration information, including 11 configuration fields.	An object of the class Recommender.
embed(data)	Generate the mapping between one-hot encoding and embedding vectors.	data: Each dimension of data is spliced by several one-hot codes. For example, the encoding of "A B-> C:" can be expressed as: [0, 0, 1, 0, 1, 0, 1, 0, 0].
most_similarity(positive, negative=None, top_n=10)	Calculate and get the most similar top_n items.	positive: List of items that contribute positively. negative: List of items that contribute negatively. top_n: Top N.	The most similar top_n items and their similarities.
infer(items)	Inferring based on the trained model.	items: Each dimension of data is spliced by several one-hot codes.	One-hot codes of desired items.
get_vector(items)	Get the embedding vectors of items.	items: one-hot codes.	Embedding vectors of items.
save(path1, path2)	Save the model and related configurations.	path1: The path to store the model. path2: The path to store the configurations.
@staticmethod load(path1, path2)	Import pre-trained models and related configurations.	path1: The path to import the model. path2: The path to import the configurations.
set_config(config)	Set the configuration information of Recommender.	config: The configuration information, including 11 configuration fields.
set_lr(learning_rate)	Set the learning rate of the neural network.	learning_rate: The learning rate of the neural network.
set_workers(workers)	Set the number of worker threads to train the model.	workers: The number of worker threads.
@staticmethod version()			Returns the version information of Recommender.

Configurable fields:

Field	Description	Default
capacity	Capacity of LycorisNet.
dimension	Dimension of each item.
sequence	The number of items.
nodes	The number of hidden nodes added for each neural network.
connections	The number of connections added for each neural network.
depths	Total layers of each neural network.
batch_size	Batch size.
epoch	Epoch.
middle_layer	Select the number of layers as the embedding vector. It starts at index 0.
evolution	Number of LycorisNet evolutions.	0
verbose	Whether to output intermediate information.	False

Usage

LycorisR is a sequence-based recommendation framework. The input data is a sequential list of items, and the framework automatically explores correlations between these items. Here is a simple example with 8 groups of sequences: A B -> C, A B -> D, B A -> C, B A -> D, C D -> A, C D -> B, D C -> A, D C -> B.

Import the dependent modules:

from LycorisR import Recommender

Prepare the data:

data = [[0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0], [0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0],
        [1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1], [1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0],
        [0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0], [0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0],
        [0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1], [0, 0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0]]

Prepare the configuration information and instantiate the Recommender object:

conf = {"capacity": 64, "dimension": 4, "sequence": 3, "nodes": 1000, "connections": 30000, "depths": 8,
        "batch_size": 8, "epoch": 16, "evolution": 8, "middle_layer": 4, "verbose": True}
lre = Recommender(conf)

Embedding:

lre.embed(data)

Inferring B C -> ?:

lre.infer([0, 0, 1, 0, 0, 1, 0, 0])

Get the most similar items to "A":

lre.most_similarity([0, 0, 0, 1])

Get the embedding vectors of "D":

lre.get_vector([1, 0, 0, 0])

More examples will be released in the future.

License

LycorisR is released under the LGPL-3.0 license. By using, distributing, or contributing to this project, you agree to the terms and conditions of this license.