fri_CS276_test_vectorial_search.py

import time
import numpy as np
import matplotlib.pyplot as plt
import fri_CS276_vectoriel_search

root_folder = "CACM/"
query_filename = "query.text"
query_answer_filename = "qrels.text"


def read_queries(filename):
    with open(filename) as f:
        queries = dict()
        to_read = (".W")
        to_ignore = (".T", ".B", ".A", ".N", ".X", ".K")
        current_marker = ""

        for line in f:
            if len(line) > 1 and line[0:2] == ".I":
                i = int(line.replace(".I", ""))
                queries[i] = {".W": "", "answers": []}
            elif len(line) > 1 and line[0:2] in to_ignore:
                current_marker = ""
            elif len(line) > 1 and line[0:2] in to_read:
                current_marker = line[0:2]
            elif current_marker:
                queries[i][current_marker] += line
    return queries

def calculate_number_of_relevant_documents(guess, expected):
    if len(expected) == 0:
        return -1
    count = 0
    for element in expected:
        if element in guess:
            count += 1
    return count


# E-Measure (van Rijsbergen)
def calculate_e_measure(precision_list, rappel_list):
    output = []
    for i in range(len(precision_list)):
        P = precision_list[i]
        R = rappel_list[i]
        beta = P / R
        output.append(1 - ((beta * beta + 1) * P * R) / (beta * beta * P + R))
    return np.mean(output)


# F-Measure (van Rijsbergen)
def calculate_f_measure(precision_list, rappel_list):
    output = []
    for i in range(len(precision_list)):
        P = precision_list[i]
        R = rappel_list[i]
        output.append((2 * P * R) / (P + R))
    return np.mean(output)


# Average mean precision
def calculate_average_mean_precision(precision, rappel):
    # precision[k][query] k=NUMBER_OF_RETURNED_DOCUMENTS, query = query index
    query_index = 0
    MAP = 0
    for query_index in range(len(precision[0])):
        average_precision = 0
        last_rappel = 0
        for k in range(len(precision)):
            delta_rappel = rappel[k][query_index] - last_rappel
            average_precision += precision[k][query_index] * delta_rappel
            last_rappel = rappel[k][query_index]
        MAP += average_precision
    return MAP / len(precision[0])


queries = read_queries(root_folder + query_filename)

last_time = time.time()
number_of_document, tfidf, normalization_dictionary, docs, id = fri_CS276_vectoriel_search.tf_idf()
print("Duration for building index : " + str((time.time() - last_time) * 1000) + " ms")
rappel_list = []
precision_list = []
rappel_list_full = []
precision_list_full = []
query_duration_list = []

rappel_list_temp = []
precision_list_temp = []
NUMBER_OF_RETURNED_DOCUMENTS = 20
for key, value in queries.items():
    query = value[".W"]
    last_time = time.time()
    answer = fri_CS276_vectoriel_search.process_query(query, number_of_document, tfidf, normalization_dictionary,
                                                     docs,
                                                     id, NUMBER_OF_RETURNED_DOCUMENTS)
    query_duration_list.append(time.time() - last_time)
    answer_list = []


print("Mean duration for one request: " + str(1000 * np.mean(query_duration_list)) + " ms")