Spaces:
Sleeping
Sleeping
Delete hist.txt
Browse files
hist.txt
DELETED
@@ -1,56 +0,0 @@
|
|
1 |
-
|
2 |
-
# ์ํค ์์ฝ ๊ด๋ จ
|
3 |
-
def extract_main_query(text):
|
4 |
-
sentences = re.split(r'[.?!]\s*', text)
|
5 |
-
sentences = [s.strip() for s in sentences if s.strip()]
|
6 |
-
if not sentences:
|
7 |
-
return text
|
8 |
-
last = sentences[-1]
|
9 |
-
last = re.sub(r'[^๊ฐ-ํฃa-zA-Z0-9 ]', '', last)
|
10 |
-
particles = ['์ด', '๊ฐ', '์', '๋', '์', '๋ฅผ', '์', '์์', '์๊ฒ', 'ํํ
', '๋ณด๋ค']
|
11 |
-
for p in particles:
|
12 |
-
last = re.sub(rf'\b(\w+){p}\b', r'\1', last)
|
13 |
-
return last.strip()
|
14 |
-
|
15 |
-
def get_wikipedia_summary(query):
|
16 |
-
cleaned_query = extract_main_query(query)
|
17 |
-
url = f"https://ko.wikipedia.org/api/rest_v1/page/summary/{cleaned_query}"
|
18 |
-
res = requests.get(url)
|
19 |
-
if res.status_code == 200:
|
20 |
-
return res.json().get("extract", "์์ฝ ์ ๋ณด๋ฅผ ์ฐพ์ ์ ์์ต๋๋ค.")
|
21 |
-
else:
|
22 |
-
return "์ํค๋ฐฑ๊ณผ์์ ์ ๋ณด๋ฅผ ๊ฐ์ ธ์ฌ ์ ์์ต๋๋ค."
|
23 |
-
|
24 |
-
def textrank_summarize(text, top_n=3):
|
25 |
-
sentences = re.split(r'(?<=[.!?])\s+', text.strip())
|
26 |
-
sentences = [s.strip() for s in sentences if len(s.strip()) > 10]
|
27 |
-
if len(sentences) <= top_n:
|
28 |
-
return text
|
29 |
-
vectorizer = TfidfVectorizer()
|
30 |
-
tfidf_matrix = vectorizer.fit_transform(sentences)
|
31 |
-
sim_matrix = cosine_similarity(tfidf_matrix)
|
32 |
-
np.fill_diagonal(sim_matrix, 0)
|
33 |
-
def pagerank(matrix, damping=0.85, max_iter=100, tol=1e-4):
|
34 |
-
N = matrix.shape[0]
|
35 |
-
ranks = np.ones(N) / N
|
36 |
-
row_sums = np.sum(matrix, axis=1)
|
37 |
-
row_sums[row_sums == 0] = 1
|
38 |
-
for _ in range(max_iter):
|
39 |
-
prev_ranks = ranks.copy()
|
40 |
-
for i in range(N):
|
41 |
-
incoming = matrix[:, i]
|
42 |
-
ranks[i] = (1 - damping) / N + damping * np.sum(incoming * prev_ranks / row_sums)
|
43 |
-
if np.linalg.norm(ranks - prev_ranks) < tol:
|
44 |
-
break
|
45 |
-
return ranks
|
46 |
-
scores = pagerank(sim_matrix)
|
47 |
-
ranked_idx = np.argsort(scores)[::-1]
|
48 |
-
selected_idx = sorted(ranked_idx[:top_n])
|
49 |
-
summary = ' '.join([sentences[i] for i in selected_idx])
|
50 |
-
return summary
|
51 |
-
|
52 |
-
def summarize_from_wikipedia(query, top_n=3):
|
53 |
-
raw_summary = get_wikipedia_summary(query)
|
54 |
-
first_summary = textrank_summarize(raw_summary, top_n=top_n)
|
55 |
-
second_summary = textrank_summarize(first_summary, top_n=top_n)
|
56 |
-
return second_summary
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|