def build_topology(matrix_S, matrix_D, active_user, target_song):
"""
Function that builds the topology of the network.
Parameters
----------
matrix_S: pd.DataFrame
matrix_D: pd.DataFrame
active_user: int
Active user ID
target_song: int
Target song ID
"""
# 1) *********************** STATIC TOPOLOGY ***********************
# 1.a) ********************* Content based *************************
# 1.a.1) Instantiate Acb.
a_cb = User(active_user, cb=True)
# 1.a.2) Instantiate songs reviewed by the active user
item_nodes = [Item(i) for i in matrix_S.columns[1:]]
# 1.a.3) Instantiate edges from items rated by active user to A_cb (i --> A_cb)
i_acb_edges = [Edge(i, a_cb) for i in item_nodes]
# 1.a.4) Instantiate features and the corresponding edges (f --> i) to their children items
feature_nodes = [Feature(f) for f in matrix_D.columns[1:]]
f_i_edges = get_edges(feature_nodes, item_nodes, matrix_D)
# 1.b) ********************* Collaborative component *************************
# 1.b.1) Instantiate Acf. Instantiate k most-similar users.
a_cf = User(active_user, cf=True)
user_nodes = get_users(active_user, matrix_S, k=4)
# 1.b.2) Instantiate edges from k-most similar users to Acf.
u_acf_edges = [Edge(u, a_cf) for u in user_nodes]
# 1.c) ********************** Hybrid Component **************************+
# 1.c.1) Instantiate Ah, create edges with Acf and Acb
a_h = User(active_user, h=True)
acf_ah_edge = Edge(a_cf, a_h)
acb_ah_edge = Edge(a_cb, a_h)
# 2) *********************** DYNAMIC TOPOLOGY ***********************
# 2.a) ********************* Content based *************************
# 2.a.1) Set target item
target_song_node = [i for i in item_nodes if i.index == target_song][0]
target_song_node.set_as_target()
# 2.a.2) Instantiate edges from all features describing target item to target item.
f_target_song_edges = get_edges(feature_nodes, [target_song_node],
matrix_D)
# 2.b) ********************* Collaborative component *************************
# 2.b.1) From the set of k-most similar users, get those that didn't rate the target item, U_.
u_minus = get_u_minus([u for u in user_nodes if u.index != active_user],
target_song_node, matrix_S)
# 2.b.2) Instantiate edges from items rated by users in U_ to users in U_.
i_u_minus_edges = get_edges(item_nodes, u_minus, matrix_S)
# 3) Return graph
nodes = [obj for obj in gc.get_objects() if isinstance(obj, Node)]
edges = [obj for obj in gc.get_objects() if isinstance(obj, Edge)]
graph = Graph(nodes, edges)
return graph
