Development - Advanced, exercise 16

Text

Co-citation is a similarity measure that uses citation analysis to establish a similarity relationship between documents. Co-citation is defined as the frequency with which two documents are cited together by other documents. The co-citation index of two given documents is higher depending on how many documents cite them both. For instance, suppose that document A cites documents D, E, F, G, document B cites documents F, H, I, and document C cites documents D, F. Thus, documents D and F have a co-citation index of 2, since both are cited together by documents A and C. Write an algorithm in Python – def co_citation(doc_1, doc_2, list_of_docs) – which returns the co-citation index of the input documents doc_1 and doc_2 calculated from documents in the input list, weighted on the total number of documents available in the input list. Each item in the input list, representing a document, is defined as the set of the documents (identified by strings) it cites. For instance, if we call the function with input documents "A" and "C" and list of documents [{"A", "B", "C"}, {"B", "D"}, {"A", "C", "E"}], it will return 0.66 as result, since the input documents are cited together in the 1st and 3rd documents in the input list and the overall number of documents in the input list is 3 – thus the weighted co-citation index is 2/3, which is equal to 0.66.

Solution

# Test case for the function
def test_co_citation(doc_1, doc_2, list_of_docs, expected):
    result = co_citation(doc_1, doc_2, list_of_docs)
    if expected == result:
        return True
    else:
        return False


# Code of the function
def co_citation(doc_1, doc_2, list_of_docs):
    if list_of_docs:
        co_citation_index = 0

        for doc in list_of_docs:
            if doc_1 in doc and doc_2 in doc:
                co_citation_index += 1

        return co_citation_index / len(list_of_docs)

    return 0


# Tests
print(test_co_citation("A", "C", [{"A", "B", "C"}, {"B", "D"}, {"A", "C", "E"}], 2 / 3))
print(test_co_citation("A", "B", [{"A", "B", "C"}, {"B", "D"}, {"A", "C", "E"}], 1 / 3))
print(test_co_citation("C", "F", [{"A", "B", "C"}, {"B", "D"}, {"A", "C", "E"}], 0))

Additional material

The runnable Python file is available online.