Use cases

Improve your products with Upgini

Detecting fake review

Auto moderation

Price estimation
for advertisers

Salary prediction
for job seekers

Protection against
a competitor's attack

Protection against
non-pickup orders

Detecting fake reviews

Detecting fake reviews

Increase sales
Proven commercial effect
Free to test
Why does it matter?
  1. Fake reviews can mislead consumers.
  2. Fake reviews can harm the reputation of your business.
  3. If fake reviews go undetected, it can erode trust in your service.
How can you solve this problem?
Create an ML application that will check every review and reviewer profile. The main task of this application is to block fake reviews. The target accuracy for such a system is 95%.
Have you experienced fake reviews?
Get know how Upgini could help you.
  1. Upgini lets you achieve best-in-class accuracy of fake review detection due to independent benchmarks using large language models (LLMs), GraphNNs and recurrent neural networks (RNNs)
  2. Upgini enriches input text reviews with contextual information from external data sources and generate more accurate features that you could generate by yourself
  3. Upgini provides automated feature generation from the text of reviews: you could collect 200+ extra features with good fake detection power.
  4. Upgini enriches your reviewer profile with new external data: more than 7 000 features available
An article with an example of using Upgini for Amazon reviews classification
https://medium.com/towards-data-science/automated-feature-engineering-in-python-5733426530bf
Try to test Upgini for fake detection
Just copy the code and send it to your analyst to repeat!

Upgini will quickly generate a quality report.Ask your analyst to send it to you to check the result.

Copied
%pip install -Uq upgini

# read labeled data
df = pd.read_csv("/content/Reviews.csv")

from upgini import FeaturesEnricher, SearchKey

# map data
enricher = FeaturesEnricher(
    search_keys={'Date': SearchKey.DATE,
                'IP_address': SearchKey.IP
                }, 
    generate_features=['text_review']
    )

# launch fit step
enricher.fit(df[['Date','IP_address','text_review']], 
             df['fraud_label'])

# launch enrichment step
df_enriched = enricher.transform(df)
Auto Moderation

Auto Moderation

of product descriptions / ad texts / CVs
Increase sales
Increase Trust&Safety
Free to test
Why does it matter?
  1. An auto moderation system is of great importance in ensuring trust and safety in online platforms
  2. The auto moderation system makes mistakes
  3. Improving quality of your Auto Moderation system let you make your Trust&Safety better
  4. Upgini has proven track record of increasing accuracy of moderation systems in the biggest online platforms
How could Upgini help you?
  1. Upgini lets you achieve best-in-class accuracy of auto moderation using large language models (LLMs), GraphNNs and recurrent neural networks (RNNs)
  2. Upgini enriches input text descriptions with contextual information from external data sources and generate more accurate features that you could generate by yourself
  3. Upgini will provide to your DWH 200+ new relevant features for auto moderation task
Price estimation

Price estimation

for advertising a product/service
Increase DAU, MAU
Proven commercial effect
Free to test
Why does it matter?
  1. The user who posts an advertisement may incorrectly evaluate the cost of their product or service
  2. Provide the user with an automatic assessment of their product or service based on the texts of the advertisement, and next time he will come back to you
How can you solve this problem?
Develop a service that determines the optimal price based on advertisement texts
How could Upgini help you?
  1. Upgini lets you achieve best-in-class accuracy for price optimization based on feature extraction from the texts due to independent benchmarks using large language models (LLMs), GraphNNs, and recurrent neural networks (RNNs).
  2. Upgini enriches input texts with contextual information from external data sources and generates more accurate features than you could generate yourself.
  3. Upgini provides automated feature generation from review text: you can collect 200+ additional features with good price estimation power.
  4. Upgini enriches your reviewer profile with new external data: more than 7,000 features are available.
Salary prediction

Salary prediction

Increase DAU, MAU
Proven commercial effect
Free to test
Why does it matter?
  1. The user who posts a resume on the site for job seekers may incorrectly evaluate the cost of their labour
  2. Provide the user with an automatic salary assessment based on the texts of the resume, and next time he will come back to your portal agian
How can you solve this problem?
Develop an ML application that will determine the fair wage of a candidate based on the resume text, not just the job position name.
How could Upgini help you?
  1. Upgini lets you achieve best-in-class accuracy for salary estimation based on feature extraction from the resume texts due to independent benchmarks using large language models (LLMs), GraphNNs, and recurrent neural networks (RNNs).
  2. Upgini enriches input resume texts with contextual information from external data sources and generates more accurate features than you could generate yourself.
  3. Upgini provides automated feature generation from the resume text: you can collect 200+ additional features with good salary prediction power.
  4. Upgini enriches your reviewer profile with new external data: more than 7,000 features are available.
Protection against a competitor's attack

Protection against a competitor's attack

Increase sales
Free testing
Proven commercial effect
Why does it matter?
There are several methods that competitors may use to attack each other:
  1. Manipulating reviews
  2. Manipulating ordering
These attacks can result in a drop in sales.
How can you solve this problem?
Develop an AI service that will check every order and every review for competitor attack.
How could Upgini help you?
Upgini lets you enrich your systems with new relevant external data that may help you to identify competitor attack by IP-address, user geo position or email: more than 7,000 features are available.
Non-pickup orders protection

Non-pickup orders protection

Increase sale
Proven commercial effect
Free to test
Why does it matter?
Orders without pickup occupy a significant share of orders in online stores.
Because of this, the store incurs extra logistics costs, does not receive profit from potential sales of blocked goods.
How can you solve this problem?
Score every postpaid order on the probability of pick-up. For orders with a high probability of non-pickup, require prepaid.
How could Upgini help you?
Upgini lets you enrich your systems with new relevant external data that may help you score orders by IP address, user geo position, email, or phone. More than 15,000 features are available to predict whether a customer will pick up an order or not.

Make ML&AI models happier
with more relevant data

No credit card required. No time limit on Free plan

Get started
Use cases
Fraud preventionMarketing and CRMDemand forecastProduct managementPersonalizationRisk managementLead scoring
Resources
DocsTutorials & Publications
PricingSell your data
for AI & ML applications
aicpa soc
SOC 2
© 2022-2024 Upgini  ·  Designed in UAE, build worldwide