Python3 - Grouping similar strings together

Question

What I'm looking to do is group strings together off of a fiction website. The titles of the posts are generally in the format something like:

titles = ['Series Name: Part 1 - This is the chapter name',
    '[OC] Series Name - Part 2 - Another name with the word chapter and extra oc at the start',
    "[OC] Series Name = part 3 = punctuation could be not matching, so we can't always trust common substrings",
    '{OC} Another cool story - Part I - This is the chapter name',
    '{OC} another cool story: part II: another post title',
    '{OC} another cool story part III but the author forgot delimiters',
    "this is a one-off story, so it doesn't have any friends"]

Delimiters etc aren't always there, and there can be some variation.

I'd start by normalizing the string to just alphanumeric characters.

import re
from pprint import pprint as pp

titles = []  # from above

normalized = []
for title in titles:
    title = re.sub(r'\bOC\b', '', title)
    title = re.sub(r'[^a-zA-Z0-9\']+', ' ', title)
    title = title.strip()
    normalized.append(title)

pp(normalized)

which gives

   ['Series Name Part 1 This is the chapter name',
 'Series Name Part 2 Another name with the word chapter and extra oc at the start',
 "Series Name part 3 punctuation could be not matching so we can't always trust common substrings",
 'Another cool story Part I This is the chapter name',
 'another cool story part II another post title',
 'another cool story part III but the author forgot delimiters',
 "this is a one off story so it doesn't have any friends"]

The output I'm hoping for is:

['Series Name', 
'Another cool story', 
"this is a one-off story, so it doesn't have any friends"]  # last element optional

I know of a few different ways to compare strings...

difflib.SequenceMatcher.ratio()

Levenshtein edit distance

I've also heard of Jaro-Winkler and FuzzyWuzzy.

But all that really matters is that we can get a number showing the similarity between the strings.

I'm thinking I need to come up with (most of) a 2D matrix comparing each string to each other. But once I've got that, I can't wrap my head around how to actually separate them into groups.

I found another post that seems to have done the first part... but then I'm not sure how to continue from there.

scipy.cluster looked promising at first... but then I was in way over my head.

Another thought was somehow combining itertools.combinations() with functools.reduce() with one of the above distance metrics.

Am I way overthinking things? It seems like this should be simple, but it's just not making sense in my head.

Answer 1

This is an implementation of the ideas put forth in CKM's answer: https://stackoverflow.com/a/61671971/42346

First take out the punctuation -- it's not important to your purpose -- using this answer: https://stackoverflow.com/a/15555162/42346

Then we'll use one of the techniques described here: https://blog.eduonix.com/artificial-intelligence/clustering-similar-sentences-together-using-machine-learning/ to cluster similar sentences.

from nltk.tokenize import RegexpTokenizer

tokenizer = RegexpTokenizer(r'\w+') # only alphanumeric characters

lol_tokenized = []
for title in titles:
    lol_tokenized.append(tokenizer.tokenize(title))

Then get a numeric representation of your titles:

import numpy as np 
from gensim.models import Word2Vec

m = Word2Vec(lol_tokenized,size=50,min_count=1,cbow_mean=1)  
def vectorizer(sent,m): 
    vec = [] 
    numw = 0 
    for w in sent: 
        try: 
            if numw == 0: 
                vec = m[w] 
            else: 
                vec = np.add(vec, m[w]) 
            numw += 1 
        except Exception as e: 
            print(e) 
    return np.asarray(vec) / numw 

l = []
for i in lol_tokenized:
    l.append(vectorizer(i,m))

X = np.array(l)

Whoo-boy that was a lot.
Now you have to do the clustering.

from sklearn.cluster import KMeans

clf = KMeans(n_clusters=2,init='k-means++',n_init=100,random_state=0)
labels = clf.fit_predict(X)
print(labels)
for index, sentence in enumerate(lol_tokenized):
    print(str(labels[index]) + ":" + str(sentence))

[1 1 0 1 0 0 0]
1:['Series', 'Name', 'Part', '1', 'This', 'is', 'the', 'chapter', 'name']
1:['OC', 'Series', 'Name', 'Part', '2', 'Another', 'name', 'with', 'the', 'word', 'chapter', 'and', 'extra', 'oc', 'at', 'the', 'start']
0:['OC', 'Series', 'Name', 'part', '3', 'punctuation', 'could', 'be', 'not', 'matching', 'so', 'we', 'can', 't', 'always', 'trust', 'common', 'substrings']
1:['OC', 'Another', 'cool', 'story', 'Part', 'I', 'This', 'is', 'the', 'chapter', 'name']
0:['OC', 'another', 'cool', 'story', 'part', 'II', 'another', 'post', 'title']
0:['OC', 'another', 'cool', 'story', 'part', 'III', 'but', 'the', 'author', 'forgot', 'delimiters']
0:['this', 'is', 'a', 'one', 'off', 'story', 'so', 'it', 'doesn', 't', 'have', 'any', 'friends']

Then you can pull out the ones with index == 1:

for index, sentence in enumerate(lol_tokenized): 
    if labels[index] == 1: 
        print(sentence) 

['Series', 'Name', 'Part', '1', 'This', 'is', 'the', 'chapter', 'name']
['OC', 'Series', 'Name', 'Part', '2', 'Another', 'name', 'with', 'the', 'word', 'chapter', 'and', 'extra', 'oc', 'at', 'the', 'start']
['OC', 'Another', 'cool', 'story', 'Part', 'I', 'This', 'is', 'the', 'chapter', 'name']

Answer 2

Your task falls into what is known as semantic similarity. I propose you proceed as follows:

1. Get a mapping of your strings through Glove/Word2vec or popular BERT. This will give you a numeric representation of your titles.
2. Then perform clustering starting with scikit's k-means and then you can go for advanced methods for clustering.