Home

Text Search

Problem Statement

• For a given text, pair a most relavent text.

References

• https://www.youtube.com/watch?v=3pRgZfjTb0c
• https://drive.google.com/file/d/1ZzM3uVmR6td0JQ9QfLs7Z0zYR2F4dCln/view

Approach

• For this project, let's use IMDB dataset of movie reviews and generate search results of review.
• Our problem seeks a similarity function for given text against dataset (all reviews). This problem is called as Nearest Neighbour Search problem (latest version of ElasticSearch has already implemented Nearest Neighbour Search)
• For vectorizer, let's use Google\Tensorflow's Universal Sentence Encoder (USE) with 512 dimensions.

Requirements

• ElasticSearch database installed and running.
• Universal Sentence Encoder (USE) model downloaded and stored locally from https://tfhub.dev/google/universal-sentence-encoder-large/5
• IMDB Dataset downloaded and stored locally from https://www.kaggle.com/lakshmi25npathi/imdb-dataset-of-50k-movie-reviews
• Installs:
  pip3 install --upgrade pip
pip3 install elasticsearch
pip3 install pandas
pip3 install --upgrade --no-cache-dir tensorflow
pip3 install --upgrade tensorflow-hub

Configurations

DATA

DATA_PATH = '/Users/shiv/Documents/gitRepositories/iutils/input/data/IMDB Dataset.csv'
TEXT_COLUMN = 'review'
NUM_OF_SAMPLES = 100

DATABASE

DB_HOST_NAME = '127.0.0.1'
DB_PORT = 9201

ENCODER

ENCODER_PATH = '/Users/shiv/Documents/gitRepositories/text_search/encoders/universal-sentence-encoder-large_5'
_encoder = hub.load(ENCODER_PATH) # Load the encoder

Results

NOTE: Results depend on # of samples in dataset, current project could be improved with some pre-processing of text

Search Text = 'psychological thriller is what i like'

Results:
{'percentage_match': 73, 'text': "Well, I like to watch bad horror B-Movies, cause I think it's interesting to see stupidity and unability of creators to shoot seriously good movie"}
{'percentage_match': 70, 'text': '"The Cell" is an exotic masterpiece, a dizzying trip into not only the vast mind of a serial killer, but also into one of a very talented director'}
{'percentage_match': 67, 'text': "Average (and surprisingly tame) Fulci giallo which means it's still quite bad by normal standards, but redeemed by its solid build-up and some nice touches such as a neat time twist on the issues of visions and clairvoyance"}
{'percentage_match': 67, 'text': 'Taut and organically gripping, Edward Dmytryk's Crossfire is a distinctive suspense thriller, an unlikely "message" movie using the look and devices of the noir cycle'}
{'percentage_match': 65, 'text': 'This movie struck home for me'}
{'percentage_match': 65, 'text': 'How this film could be classified as Drama, I have no idea'}
{'percentage_match': 64, 'text': 'This film took me by surprise'}
{'percentage_match': 64, 'text': 'This film laboured along with some of the most predictable story lines and shallow characters ever seen'}
{'percentage_match': 63, 'text': 'Oh noes one of these attack of the Japanese ghost girl movies'}
{'percentage_match': 63, 'text': 'I really like Salman Kahn so I was really disappointed when I seen this movie'}