Assigning roles using spaCy's parser

Assigning roles using spaCy's parser

In this exercise you'll use spaCy's powerful syntax parser to assign roles to the entities in your users' messages. To do this, you'll define two functions, find_parent_item() and assign_colors(). In doing so, you'll use a parse tree to assign roles, similar to how Alan did in the video.

Recall that you can access the ancestors of a word using its .ancestors attribute.

Instructions

  • Create a spacy document called doc by passing the message "let's see that jacket in red and some blue jeans" to the nlp object.
  • In the find_parent_item(word) function, iterate over the ancestors of each word until an entity_type() of "item" is found.
  • In the assign_colors(doc) function, iterate over the doc until an entity_type of "color" is found. Then, find the parent item of this word.
  • Pass in the spacy document to the assign_colors() function.

====================================================================

# Create the document
doc = nlp("let's see that jacket in red and some blue jeans")

# Iterate over parents in parse tree until an item entity is found
def find_parent_item(word):
    # Iterate over the word's ancestors
    for parent in word.ancestors:
        # Check for an "item" entity
        if entity_type(parent) == "item":
            return parent.text
    return None

# For all color entities, find their parent item
def assign_colors(doc):
    # Iterate over the document
    for word in doc:
        # Check for "color" entities
        if entity_type(word) == "color":
            # Find the parent
            item =  find_parent_item(word)
            print("item: {0} has color : {1}".format(item, word))

# Assign the colors
assign_colors(doc)

Last modified: Tuesday, 21 February 2023, 10:11 AM