RecSummarizer

A framework to generate summaries as explanations for recommendation engines.

Instalation

Install our library:

pip install git+https://github.com/luanssouza/recsummarizer

Install Stanza:

pip install stanza

Example Usage

Obtain our reference corpus:

wget https://raw.githubusercontent.com/luanssouza/recsummarizer/main/resources/BNC_nouns.csv

Import the modules:

# External modules
import stanza
import pandas as pd
# Our modules
from recsummarizer.corpus import CsvGeneralCorpus
from recsummarizer.normalize.tf_idf_normalizer import TfIdfNormalizer
from recsummarizer.centroid.centroid import Centroid
from recsummarizer.persistence.stanza_persistence import StanzaPersistence
from recsummarizer.preprocess.stanza_preprocess import StanzaPreProcess
from recsummarizer.summarize.summarizer_baseline import SummarizerBaseline
from recsummarizer.embedding.bert_embedding import BertEmbedding

Initialize the models:

# Downloading english model
stanza.download('en', verbose=False)

# Creating our pipeline
nlp = stanza.Pipeline('en', processors='tokenize,mwt,pos,sentiment', verbose=False)

embedding = BertEmbedding('all-MiniLM-L6-v2')

general_corpus = CsvGeneralCorpus(pd.read_csv('./BNC_nouns.csv', index_col='noun'))

Preprocess the reviews:

raw_reviews = [
    "The director is really good. The movie is awesome! You will definetly enjoy it! The scenes are the best!", 
    "The racer is really good."
]

items = [
    {"id": 0, "reviews": raw_reviews }
]

# Creating normalizer instance
normalizer = TfIdfNormalizer()

# Creating centroid instance
centroid = Centroid(normalizer, 0.35)

# Creating persistence instance
persistence = StanzaPersistence('./data/', embedding, centroid)

# Creating a instance of PreProcess
preprocess = StanzaPreProcess(-20, 20)

# Preprocessing movies
preprocess.proprocess(items, persistence, general_corpus, nlp)

Obtain the summary:

SummarizerBaseline('./data/', 0.90, 5).summarize(0)

Cite Us

@inproceedings{10.1145/3539637.3557002,
    author = {Souza, Luan Soares de and Manzato, Marcelo Garcia},
    title = {Aspect-Based Summarization: An Approach With Different Levels of Details to Explain Recommendations},
    year = {2022},
    isbn = {9781450394093},
    publisher = {Association for Computing Machinery},
    address = {New York, NY, USA},
    url = {https://doi.org/10.1145/3539637.3557002},
    doi = {10.1145/3539637.3557002},
    booktitle = {Proceedings of the Brazilian Symposium on Multimedia and the Web},
    pages = {202–210},
    numpages = {9},
    keywords = {hierarchical clustering, multi-level summarization},
    location = {Curitiba, Brazil},
    series = {WebMedia '22}
}