The file main.py represents the process of finding metaphors in a text:

1. Text Segmentation
2. Finding candidates for the metaphors: a candidate is a pair of word: adjective-noun or verb-noun
3. Labeling the metaphors: Is a candidate metaphorical or literal?

Arguments:

• -ml or --labeler followed by an ID: choose the metaphor labeling method:
  • darkthoughts
  • cluster
  • kmeans
  • Default value: darkthoughts
• -cf or --finder followed by an ID: choose the candidate finding method:
  • adjNoun
  • verbNoun
  • Default value: adjNoun
• -v or --verbose
  • Print the different steps of the process
  • Default value: False
• -f or --file followed by a file name: look for metaphors in a text file
• -s or --string followed by a string: look for metaphors in a specified string
• -cg or --cgenerator:
  • Useful when combined with an excel or csv file. Use word pair in the file as candidates instead of looking for candidates in the annotated text
  • Default value: False

If no string or text file is specified in the command line then a default text is used.

1. Parsing the command line
2. Creating a hash-table
   1. Adding the candidate finder functions
   2. Adding the metaphor labeler functions
3. Initializing the text either from:
   1. Default text - defined in modules/utils.py
   2. A string - written in the command line
   3. A file - path in the command line
4. Creating the object MetaphorIdentification

The AnnotatedText is created from the raw text using the nltk.word_tokenize() function. The Part-of-Speech and the Lemma of each word is also determined with NLTK functions: nltk.pos_tag and nltk.WordNetLemmatizer.

Call the procedure MetaphorIdentification.findCandidates()

Call the procedure MetaphorIdentification.labelMetaphors()

Defined in /new_structure/modules/datastructs/registry.py.

To identify metaphors in a text, at least two steps need to be followed: the candidate identification step and the labelling step. Each of these steps can be done in many ways. Each method needs to be registered in the metaphorRegistry defined in /sample/modules/registry.py

Defined in /new_structure/modules/datastructs/MetaphorIdentification.py.

It has four fields:

• rawText: string
• annotatedText: class AnnotatedText from modules/datastructs/annotated_text.py
• candidates: class CandidateGroup from modules/datastructs/candidate_group.py
• metaphors: class MetaphorGroup from modules/datastructs/labeled_metaphor_list.py

##How to Add a New Metaphor-Labeling Function Your function must be defined in a new file in the modules folder.

The input of the function must be:

• candidates
  • Type: Object of class CandidateGroup
• cand_type:
  • Type: string
  • Value: "adjNoun" or "verbNoun"
  • Usage: Corresponds to a database
• verbose:
  • Type: Boolean
  • Usage: Display some information if its value is True

The output of the function must be an object of class MetaphorGroup

• Variables
  • candidates: list of objects of class Candidate
  • size: number of elements in the list above
• Methods
  • addCandidate(candidate): Add the element candidate to the list candidates and increment the variable size by 1
  • getCandidate(index): Return the candidate of index index in the list candidates
  • __iter__()
  • __str__()

• Variables
  • metaphors: list of objects of class Metaphor
  • size: number of elements in the list above
• Methods
  • addMetaphor(metaphor): Add the element metaphor to the list metaphors and increment the variable size by 1
  • getMetaphor(index): Return the metaphor of index index in the list metaphors
  • writeToCSV()
  • __iter__()
  • __str__()

• Variables
  • annotatedText: object of class AnnotatedText
  • sourceIndex: index of the source in the annotatedText
  • sourceSpan: 2-tuple = (index of the first word in the source, index of the last word in the source)
  • targetIndex: index of the target in the annotatedText
  • targetSpan: 2-tuple = (index of the first word in the target , index of the last word in the target)
• Methods
  • getSource(): return the first word of the source
  • getTarget(): return the first word of the target
  • getFullSource()
  • getFullTarget()
  • __stringAdder(): used in the getFull... functions

• Variables
  • candidate: object of class candidate
  • result: boolean
  • confidence: number between 0 and 1
• Methods
  • getSource(): return candidate.getFullSource()
  • getTarget(): return candidate.getFullTarget()
  • getResult()
  • getConfidence()
  • __str__()