Obsei is an automation tool for text analysis need
Project description
Note: Obsei is still in alpha stage hence carefully use it in Production. Also, as it is constantly undergoing development hence master branch may contain many breaking changes. Please use released version.
Obsei (pronounced "Ob see" | /əb-'sē/) is an open-source, low-code, AI powered automation tool. Obsei consists of -
- Observer: Collect unstructured data from various sources like tweets from Twitter, Subreddit comments on Reddit, page post's comments from Facebook, App Stores reviews, Google reviews, Amazon reviews, News, Website, etc.
- Analyzer: Analyze unstructured data collected with various AI tasks like classification, sentiment analysis, translation, PII, etc.
- Informer: Send analyzed data to various destinations like ticketing platforms, data storage, dataframe, etc so that the user can take further actions and perform analysis on the data.
All the Observers can store their state in databases (Sqlite, Postgres, MySQL, etc.), making Obsei suitable for scheduled jobs or serverless applications.
Future direction -
- Text, Image, Audio, Documents and Video oriented workflows
- Collect data from every possible private and public channels
- Add every possible workflow to an AI downstream application to automate manual cognitive workflows
Introduction and demo video
Use cases
Obsei use cases are following, but not limited to -
- Social listening: Listening about social media posts, comments, customer feedback, etc.
- Alerting/Notification: To get auto-alerts for events such as customer complaints, qualified sales leads, etc.
- Automatic customer issue creation based on customer complaints on Social Media, Email, etc.
- Automatic assignment of proper tags to tickets based content of customer complaint for example login issue, sign up issue, delivery issue, etc.
- Extraction of deeper insight from feedbacks on various platforms
- Market research
- Creation of dataset for various AI tasks
- Many more based on creativity 💡
Companies/Projects using Obsei
Here are some companies/projects (alphabetical order) using Obsei. To add your company/project to the list, please raise a PR or contact us via email.
- 1Page: Giving a better context in meetings and calls
- Spacepulse: The operating system for spaces
- Superblog: A blazing fast alternative to WordPress and Medium
- Zolve: Creating a financial world beyond borders
- Utilize: No-code app builder for businesses with a deskless workforce
Tutorials
💡Tips: Handle large text classification via Obsei
Demo
We have a minimal streamlit based UI that you can use to test Obsei.
Watch UI demo video
Check demo at or at Demo Link
(Note: Sometimes the Streamlit demo might not work due to rate limiting, use the docker image (locally) in such cases.)
To test locally, just run
docker run -d --name obesi-ui -p 8501:8501 obsei/obsei-ui-demo
# You can find the UI at http://localhost:8501
To run Obsei workflow easily using GitHub Actions (no sign ups and cloud hosting required), refer to this repo.
Documentation
For detailed installation instructions, usages and examples, refer to our documentation.
Support and Release Matrix
Linux | Mac | Windows1 | Remark | |
---|---|---|---|---|
Tests | ✅ | ✅ | ✅ | Low Coverage as difficult to test 3rd party libs |
PIP | ✅ | ✅ | ✅ | Fully Supported |
Conda2 | ✅ | ✅ | ✅ | Partially Supported |
1. On Windows you have to install pytorch manually. Refer to the Pytorch official instruction. ↩
2. Conda channel is missing few dependencies, hence install missing dependencies manually - ↩
Missing Conda dependencies -
pip install presidio-analyzer
pip install presidio-anonymizer
pip install zenpy
pip install searchtweets-v2
pip install google-play-scraper
pip install tweet-preprocessor
pip install gnews
pip install python-facebook-api
# GPL dependency
pip install trafilatura
How to use
Expand the following steps and create your workflow -
Step 1: Prerequisite
Install the following (if not present already) -
- Install Python 3.7+
- Install PIP (Optional if you prefer Conda)
- Install Conda (Optional if you prefer PIP)
Step 2: Install Obsei
You can install Obsei either via PIP or Conda based on your preference.
NOTE: On Windows you have to install pytorch manually. Refer to https://pytorch.org/get-started/locally/.
Install via PIP:
To install latest released version -
pip install obsei
Install from master branch (if you want to try the latest features) -
git clone https://github.com/obsei/obsei.git
cd obsei
pip install --editable .
Install via Conda:
To install the latest version -
conda install -c lalitpagaria obsei
Install from master branch (if you want to try the latest features) -
git clone https://github.com/obsei/obsei.git
cd obsei
conda env create -f conda/environment.yml
For GPU based local environment -
git clone https://github.com/obsei/obsei.git
cd obsei
conda env create -f conda/gpu-environment.yml
Step 3: Configure Source/Observer
from obsei.source.twitter_source import TwitterCredentials, TwitterSource, TwitterSourceConfig
# initialize twitter source config
source_config = TwitterSourceConfig(
keywords=["issue"], # Keywords, @user or #hashtags
lookup_period="1h", # Lookup period from current time, format: `<number><d|h|m>` (day|hour|minute)
cred_info=TwitterCredentials(
# Enter your twitter consumer key and secret. Get it from https://developer.twitter.com/en/apply-for-access
consumer_key="<twitter_consumer_key>",
consumer_secret="<twitter_consumer_secret>",
bearer_token='<ENTER BEARER TOKEN>',
)
)
# initialize tweets retriever
source = TwitterSource()
|
Youtube Scrapperfrom obsei.source.youtube_scrapper import YoutubeScrapperSource, YoutubeScrapperConfig
# initialize Youtube source config
source_config = YoutubeScrapperConfig(
video_url="https://www.youtube.com/watch?v=uZfns0JIlFk", # Youtube video URL
fetch_replies=True, # Fetch replies to comments
max_comments=10, # Total number of comments and replies to fetch
lookup_period="1Y", # Lookup period from current time, format: `<number><d|h|m|M|Y>` (day|hour|minute|month|year)
)
# initialize Youtube comments retriever
source = YoutubeScrapperSource()
|
from obsei.source.facebook_source import FacebookCredentials, FacebookSource, FacebookSourceConfig
# initialize facebook source config
source_config = FacebookSourceConfig(
page_id="110844591144719", # Facebook page id, for example this one for Obsei
lookup_period="1h", # Lookup period from current time, format: `<number><d|h|m>` (day|hour|minute)
cred_info=FacebookCredentials(
# Enter your facebook app_id, app_secret and long_term_token. Get it from https://developers.facebook.com/apps/
app_id="<facebook_app_id>",
app_secret="<facebook_app_secret>",
long_term_token="<facebook_long_term_token>",
)
)
# initialize facebook post comments retriever
source = FacebookSource()
|
from obsei.source.email_source import EmailConfig, EmailCredInfo, EmailSource
# initialize email source config
source_config = EmailConfig(
# List of IMAP servers for most commonly used email providers
# https://www.systoolsgroup.com/imap/
# Also, if you're using a Gmail account then make sure you allow less secure apps on your account -
# https://myaccount.google.com/lesssecureapps?pli=1
# Also enable IMAP access -
# https://mail.google.com/mail/u/0/#settings/fwdandpop
imap_server="imap.gmail.com", # Enter IMAP server
cred_info=EmailCredInfo(
# Enter your email account username and password
username="<email_username>",
password="<email_password>"
),
lookup_period="1h" # Lookup period from current time, format: `<number><d|h|m>` (day|hour|minute)
)
# initialize email retriever
source = EmailSource()
|
Google Maps Reviews Scrapperfrom obsei.source import OSGoogleMapsReviewsSource, OSGoogleMapsReviewsConfig
# initialize Outscrapper Maps review source config
source_config = OSGoogleMapsReviewsConfig(
# Collect API key from https://outscraper.com/
api_key="<Enter Your API Key>",
# Enter Google Maps link or place id
# For example below is for the "Taj Mahal"
queries=["https://www.google.co.in/maps/place/Taj+Mahal/@27.1751496,78.0399535,17z/data=!4m5!3m4!1s0x39747121d702ff6d:0xdd2ae4803f767dde!8m2!3d27.1751448!4d78.0421422"],
number_of_reviews=10,
)
# initialize Outscrapper Maps review retriever
source = OSGoogleMapsReviewsSource()
|
AppStore Reviews Scrapperfrom obsei.source.appstore_scrapper import AppStoreScrapperConfig, AppStoreScrapperSource
# initialize app store source config
source_config = AppStoreScrapperConfig(
# Need two parameters app_id and country.
# `app_id` can be found at the end of the url of app in app store.
# For example - https://apps.apple.com/us/app/xcode/id497799835
# `310633997` is the app_id for xcode and `us` is country.
countries=["us"],
app_id="310633997",
lookup_period="1h" # Lookup period from current time, format: `<number><d|h|m>` (day|hour|minute)
)
# initialize app store reviews retriever
source = AppStoreScrapperSource()
|
Play Store Reviews Scrapperfrom obsei.source.playstore_scrapper import PlayStoreScrapperConfig, PlayStoreScrapperSource
# initialize play store source config
source_config = PlayStoreScrapperConfig(
# Need two parameters package_name and country.
# `package_name` can be found at the end of the url of app in play store.
# For example - https://play.google.com/store/apps/details?id=com.google.android.gm&hl=en&gl=US
# `com.google.android.gm` is the package_name for xcode and `us` is country.
countries=["us"],
package_name="com.google.android.gm",
lookup_period="1h" # Lookup period from current time, format: `<number><d|h|m>` (day|hour|minute)
)
# initialize play store reviews retriever
source = PlayStoreScrapperSource()
|
from obsei.source.reddit_source import RedditConfig, RedditSource, RedditCredInfo
# initialize reddit source config
source_config = RedditConfig(
subreddits=["wallstreetbets"], # List of subreddits
# Reddit account username and password
# You can also enter reddit client_id and client_secret or refresh_token
# Create credential at https://www.reddit.com/prefs/apps
# Also refer https://praw.readthedocs.io/en/latest/getting_started/authentication.html
# Currently Password Flow, Read Only Mode and Saved Refresh Token Mode are supported
cred_info=RedditCredInfo(
username="<reddit_username>",
password="<reddit_password>"
),
lookup_period="1h" # Lookup period from current time, format: `<number><d|h|m>` (day|hour|minute)
)
# initialize reddit retriever
source = RedditSource()
|
Reddit ScrapperNote: Reddit heavily rate limit scrappers, hence use it to fetch small data during long period from obsei.source.reddit_scrapper import RedditScrapperConfig, RedditScrapperSource
# initialize reddit scrapper source config
source_config = RedditScrapperConfig(
# Reddit subreddit, search etc rss url. For proper url refer following link -
# Refer https://www.reddit.com/r/pathogendavid/comments/tv8m9/pathogendavids_guide_to_rss_and_reddit/
url="https://www.reddit.com/r/wallstreetbets/comments/.rss?sort=new",
lookup_period="1h" # Lookup period from current time, format: `<number><d|h|m>` (day|hour|minute)
)
# initialize reddit retriever
source = RedditScrapperSource()
|
Google Newsfrom obsei.source.google_news_source import GoogleNewsConfig, GoogleNewsSource
# initialize Google News source config
source_config = GoogleNewsConfig(
query='bitcoin',
max_results=5,
# To fetch full article text enable `fetch_article` flag
# By default google news gives title and highlight
fetch_article=True,
# proxy='http://127.0.0.1:8080'
)
# initialize Google News retriever
source = GoogleNewsSource()
|
Web Crawlerfrom obsei.source.website_crawler_source import TrafilaturaCrawlerConfig, TrafilaturaCrawlerSource
# initialize website crawler source config
source_config = TrafilaturaCrawlerConfig(
urls=['https://obsei.github.io/obsei/']
)
# initialize website text retriever
source = TrafilaturaCrawlerSource()
|
Pandas DataFrameimport pandas as pd
from obsei.source.pandas_source import PandasSource, PandasSourceConfig
# Initialize your Pandas DataFrame from your sources like csv, excel, sql etc
# In following example we are reading csv which have two columns title and text
csv_file = "https://raw.githubusercontent.com/deepset-ai/haystack/master/tutorials/small_generator_dataset.csv"
dataframe = pd.read_csv(csv_file)
# initialize pandas sink config
sink_config = PandasSourceConfig(
dataframe=dataframe,
include_columns=["score"],
text_columns=["name", "degree"],
)
# initialize pandas sink
sink = PandasSource()
|
Step 4: Configure Analyzer
Note: To run transformers in an offline mode, check transformers offline mode.
Some analyzer support GPU and to utilize pass device parameter. List of possible values of device parameter (default value auto):
- auto: GPU (cuda:0) will be used if available otherwise CPU will be used
- cpu: CPU will be used
- cuda:{id} - GPU will be used with provided CUDA device id
Text ClassificationText classification: Classify text into user provided categories. from obsei.analyzer.classification_analyzer import ClassificationAnalyzerConfig, ZeroShotClassificationAnalyzer
# initialize classification analyzer config
# It can also detect sentiments if "positive" and "negative" labels are added.
analyzer_config=ClassificationAnalyzerConfig(
labels=["service", "delay", "performance"],
)
# initialize classification analyzer
# For supported models refer https://huggingface.co/models?filter=zero-shot-classification
text_analyzer = ZeroShotClassificationAnalyzer(
model_name_or_path="typeform/mobilebert-uncased-mnli",
device="auto"
)
|
Sentiment AnalyzerSentiment Analyzer: Detect the sentiment of the text. Text classification can also perform sentiment analysis but if you don't want to use heavy-duty NLP model then use less resource hungry dictionary based Vader Sentiment detector. from obsei.analyzer.sentiment_analyzer import VaderSentimentAnalyzer
# Vader does not need any configuration settings
analyzer_config=None
# initialize vader sentiment analyzer
text_analyzer = VaderSentimentAnalyzer()
|
NER AnalyzerNER (Named-Entity Recognition) Analyzer: Extract information and classify named entities mentioned in text into pre-defined categories such as person names, organizations, locations, medical codes, time expressions, quantities, monetary values, percentages, etc from obsei.analyzer.ner_analyzer import NERAnalyzer
# NER analyzer does not need configuration settings
analyzer_config=None
# initialize ner analyzer
# For supported models refer https://huggingface.co/models?filter=token-classification
text_analyzer = NERAnalyzer(
model_name_or_path="elastic/distilbert-base-cased-finetuned-conll03-english",
device = "auto"
)
|
Translatorfrom obsei.analyzer.translation_analyzer import TranslationAnalyzer
# Translator does not need analyzer config
analyzer_config = None
# initialize translator
# For supported models refer https://huggingface.co/models?pipeline_tag=translation
analyzer = TranslationAnalyzer(
model_name_or_path="Helsinki-NLP/opus-mt-hi-en",
device = "auto"
)
|
PII Anonymizerfrom obsei.analyzer.pii_analyzer import PresidioEngineConfig, PresidioModelConfig, \
PresidioPIIAnalyzer, PresidioPIIAnalyzerConfig
# initialize pii analyzer's config
analyzer_config = PresidioPIIAnalyzerConfig(
# Whether to return only pii analysis or anonymize text
analyze_only=False,
# Whether to return detail information about anonymization decision
return_decision_process=True
)
# initialize pii analyzer
analyzer = PresidioPIIAnalyzer(
engine_config=PresidioEngineConfig(
# spacy and stanza nlp engines are supported
# For more info refer
# https://microsoft.github.io/presidio/analyzer/developing_recognizers/#utilize-spacy-or-stanza
nlp_engine_name="spacy",
# Update desired spacy model and language
models=[PresidioModelConfig(model_name="en_core_web_lg", lang_code="en")]
)
)
|
Dummy AnalyzerDummy Analyzer: Does nothing. Its simply used for transforming the input (TextPayload) to output (TextPayload) and adding the user supplied dummy data. from obsei.analyzer.dummy_analyzer import DummyAnalyzer, DummyAnalyzerConfig
# initialize dummy analyzer's configuration settings
analyzer_config = DummyAnalyzerConfig()
# initialize dummy analyzer
analyzer = DummyAnalyzer()
|
Step 5: Configure Sink/Informer
Slackfrom obsei.sink.slack_sink import SlackSink, SlackSinkConfig
# initialize slack sink config
sink_config = SlackSinkConfig(
# Provide slack bot/app token
# For more detail refer https://slack.com/intl/en-de/help/articles/215770388-Create-and-regenerate-API-tokens
slack_token="<Slack_app_token>",
# To get channel id refer https://stackoverflow.com/questions/40940327/what-is-the-simplest-way-to-find-a-slack-team-id-and-a-channel-id
channel_id="C01LRS6CT9Q"
)
# initialize slack sink
sink = SlackSink()
|
Zendeskfrom obsei.sink.zendesk_sink import ZendeskSink, ZendeskSinkConfig, ZendeskCredInfo
# initialize zendesk sink config
sink_config = ZendeskSinkConfig(
# provide zendesk domain
domain="zendesk.com",
# provide subdomain if you have one
subdomain=None,
# Enter zendesk user details
cred_info=ZendeskCredInfo(
email="<zendesk_user_email>",
password="<zendesk_password>"
)
)
# initialize zendesk sink
sink = ZendeskSink()
|
Jirafrom obsei.sink.jira_sink import JiraSink, JiraSinkConfig
# For testing purpose you can start jira server locally
# Refer https://developer.atlassian.com/server/framework/atlassian-sdk/atlas-run-standalone/
# initialize Jira sink config
sink_config = JiraSinkConfig(
url="http://localhost:2990/jira", # Jira server url
# Jira username & password for user who have permission to create issue
username="<username>",
password="<password>",
# Which type of issue to be created
# For more information refer https://support.atlassian.com/jira-cloud-administration/docs/what-are-issue-types/
issue_type={"name": "Task"},
# Under which project issue to be created
# For more information refer https://support.atlassian.com/jira-software-cloud/docs/what-is-a-jira-software-project/
project={"key": "CUS"},
)
# initialize Jira sink
sink = JiraSink()
|
ElasticSearchfrom obsei.sink.elasticsearch_sink import ElasticSearchSink, ElasticSearchSinkConfig
# For testing purpose you can start Elasticsearch server locally via docker
# `docker run -d --name elasticsearch -p 9200:9200 -e "discovery.type=single-node" elasticsearch:7.9.2`
# initialize Elasticsearch sink config
sink_config = ElasticSearchSinkConfig(
# Elasticsearch server hostname
host="localhost",
# Elasticsearch server port
port=9200,
# Index name, it will create if not exist
index_name="test",
)
# initialize Elasticsearch sink
sink = ElasticSearchSink()
|
Httpfrom obsei.sink.http_sink import HttpSink, HttpSinkConfig
# For testing purpose you can create mock http server via postman
# For more details refer https://learning.postman.com/docs/designing-and-developing-your-api/mocking-data/setting-up-mock/
# initialize http sink config (Currently only POST call is supported)
sink_config = HttpSinkConfig(
# provide http server url
url="https://localhost:8080/api/path",
# Here you can add headers you would like to pass with request
headers={
"Content-type": "application/json"
}
)
# To modify or converting the payload, create convertor class
# Refer obsei.sink.dailyget_sink.PayloadConvertor for example
# initialize http sink
sink = HttpSink()
|
Pandas DataFramefrom pandas import DataFrame
from obsei.sink.pandas_sink import PandasSink, PandasSinkConfig
# initialize pandas sink config
sink_config = PandasSinkConfig(
dataframe=DataFrame()
)
# initialize pandas sink
sink = PandasSink()
|
LoggerThis is useful for testing and dry running the pipeline. from obsei.sink.logger_sink import LoggerSink, LoggerSinkConfig
import logging
import sys
logger = logging.getLogger("Obsei")
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
# initialize logger sink config
sink_config = LoggerSinkConfig(
logger=logger,
level=logging.INFO
)
# initialize logger sink
sink = LoggerSink()
|
Step 6: Join and create workflow
source
will fetch data from the selected source, then feed it to the analyzer
for processing, whose output we feed into a sink
to get notified at that sink.
# Uncomment if you want logger
# import logging
# import sys
# logger = logging.getLogger(__name__)
# logging.basicConfig(stream=sys.stdout, level=logging.INFO)
# This will fetch information from configured source ie twitter, app store etc
source_response_list = source.lookup(source_config)
# Uncomment if you want to log source response
# for idx, source_response in enumerate(source_response_list):
# logger.info(f"source_response#'{idx}'='{source_response.__dict__}'")
# This will execute analyzer (Sentiment, classification etc) on source data with provided analyzer_config
analyzer_response_list = text_analyzer.analyze_input(
source_response_list=source_response_list,
analyzer_config=analyzer_config
)
# Uncomment if you want to log analyzer response
# for idx, an_response in enumerate(analyzer_response_list):
# logger.info(f"analyzer_response#'{idx}'='{an_response.__dict__}'")
# Analyzer output added to segmented_data
# Uncomment to log it
# for idx, an_response in enumerate(analyzer_response_list):
# logger.info(f"analyzed_data#'{idx}'='{an_response.segmented_data.__dict__}'")
# This will send analyzed output to configure sink ie Slack, Zendesk etc
sink_response_list = sink.send_data(analyzer_response_list, sink_config)
# Uncomment if you want to log sink response
# for sink_response in sink_response_list:
# if sink_response is not None:
# logger.info(f"sink_response='{sink_response}'")
Step 7: Execute workflow
Copy the code snippets from Steps 3 to 6 into a python file, for exampleexample.py
and execute the following command -
python example.py
Articles
Sr. No. | Title | Author |
---|---|---|
1 | AI based Comparative Customer Feedback Analysis Using Obsei | Reena Bapna |
2 | LinkedIn App - User Feedback Analysis | Himanshu Sharma |
Discussion forum
Discussion about Obsei can be done at community forum
Upcoming release and changelog
Upcoming release plan and progress can be tracked at link (Suggestions are welcome). Refer releases for changelogs.
Security Issue
For any security issue please contact us via email
Stargazers over time
Maintainer
This project is being maintained by Lalit Pagaria.
License
- Copyright holder: Lalit Pagaria
- Overall Apache 2.0 and you can read License file.
- Multiple other secondary permissive or weak copyleft licenses (LGPL, MIT, BSD etc.) for third-party components refer Attribution.
- To make project more commercial friendly, we void third party components which have strong copyleft licenses (GPL, AGPL etc.) into the project.
Attribution
This could not have been possible without these open source softwares.
Contribution
First off, thank you for even considering contributing to this package, every contribution big or small is greatly appreciated. Please refer our Contribution Guideline and Code of Conduct.
Thanks so much to all our contributors
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.