Similarity interface

Similarity (Interface)

This file provides the framework for various methods of track similarity calculations. Each similarity concept/ idea must inherent from the similarity interface.

Similarity Interface Documentation

Similarity

Bases: ABC

This class serves as an outline of the Similarity module for the MIAS application

There are multiple ways of achieving a recommendation system/ similarity of tracks. This modular build allows others to add their own classes to try out various methods in a simple way

Note: The playlist has already been added to the tracks dataset, such that features are calculated over the entire database. The playlist tracks are then removed from the tracks dataaset in the similarity calculation.

Additionally: It is recommended that your class take in both playlist and tracks as done in the Cosine Similarity Class

Source code in src/similarity_interface.py

class Similarity(ABC):
    """This class serves as an outline of the Similarity module for the MIAS application

    There are multiple ways of achieving a recommendation system/ similarity of tracks.
    This modular build allows others to add their own classes to try out various methods in
    a simple way

    Note: The playlist has already been added to the tracks dataset, such that features are
    calculated over the entire database. The playlist tracks are then removed from the tracks dataaset
    in the similarity calculation.

    Additionally: It is recommended that your class take in both `playlist` and `tracks` as done
    in the Cosine Similarity Class
    """
    @abstractmethod
    def calculate_similarity(self):
        """Method calculates the similarity of each track in the dataset to the given playlist feature vector

        Note, this method should populate a field called `self.similarity`, such that it represents a similarity score of
        each track to playlist vector. Please see `similarity.py` for an example implementation.
        """
        pass

    @abstractmethod
    def access_similarity_scores(self):
        """This method is a getter method providing access to the similarity metrics calculated in `calculate_similarity()`"""
        pass

    @abstractmethod
    def get_top_n(self, n: int):
        """This method should return the top-n most similar tracks as a Dataframe with essential features included.
        Note, see `similarity.py` for an example implementation
        Note, depending on your similarity calculation, be careful of the type of ordering you implement.
        """
        pass

    @abstractmethod
    def weight_features(self, weighted_columns: list):
        """This method allows for specific features to be weighted or have a more significant impact on the measures of similarity.
        Note, weighting may vary depending on the method of similarity calculation. Please see `similarity.py` for an example
        implementation
        """
        pass

access_similarity_scores() abstractmethod

This method is a getter method providing access to the similarity metrics calculated in calculate_similarity()

Source code in src/similarity_interface.py

@abstractmethod
def access_similarity_scores(self):
    """This method is a getter method providing access to the similarity metrics calculated in `calculate_similarity()`"""
    pass

calculate_similarity() abstractmethod

Method calculates the similarity of each track in the dataset to the given playlist feature vector

Note, this method should populate a field called self.similarity, such that it represents a similarity score of each track to playlist vector. Please see similarity.py for an example implementation.

Source code in src/similarity_interface.py

@abstractmethod
def calculate_similarity(self):
    """Method calculates the similarity of each track in the dataset to the given playlist feature vector

    Note, this method should populate a field called `self.similarity`, such that it represents a similarity score of
    each track to playlist vector. Please see `similarity.py` for an example implementation.
    """
    pass

get_top_n(n) abstractmethod

This method should return the top-n most similar tracks as a Dataframe with essential features included. Note, see similarity.py for an example implementation Note, depending on your similarity calculation, be careful of the type of ordering you implement.

Source code in src/similarity_interface.py

@abstractmethod
def get_top_n(self, n: int):
    """This method should return the top-n most similar tracks as a Dataframe with essential features included.
    Note, see `similarity.py` for an example implementation
    Note, depending on your similarity calculation, be careful of the type of ordering you implement.
    """
    pass

weight_features(weighted_columns) abstractmethod

This method allows for specific features to be weighted or have a more significant impact on the measures of similarity. Note, weighting may vary depending on the method of similarity calculation. Please see similarity.py for an example implementation

Source code in src/similarity_interface.py

@abstractmethod
def weight_features(self, weighted_columns: list):
    """This method allows for specific features to be weighted or have a more significant impact on the measures of similarity.
    Note, weighting may vary depending on the method of similarity calculation. Please see `similarity.py` for an example
    implementation
    """
    pass