RA Technology

Our LDA works great on 6M pieces of social media text (facebook, twitter, youtube...). Categorize text into topics with NO TRAINING DATA.
We hope that could be a good improvement for your search tool, vimp.co

Categorizing unlabeled documents based on Topic Modeling In the previous posts, we introduce Topic Modeling with Latent Dirichlet Allocation Model on theoretical aspect and applications in finding similar documents and similar users. In this post, we wil…

We love to help you understand your users/customers better.
Hope this service would be useful for you, Money Lover.

Money spending message classification Problem We have an interesting talk with Money Lover. It’s an money/finance management for people. Using this app, an user will note a message each time they spend money. A financial manageme…

We don't like code review much, we choose quality instead. Check how we do fast code release using A/B testing method.

Fast code release A/B testing based We are SaaS (Software as a service) and work with http requests. It’s very important to divide requests into small part and release new features on each piece. So we design a A/B testing syst…

Give us content, and we'll let you know how many people will love it.
Check our new approach of consumption prediction and user clustering problem.

Finding similar users based on content reading behaviors In our Recommender System, we have an interesting problem. User A has similar reading behavior with many users, such as user B. In other words, A tends to choose some in documents which B read. Our…

Simple and effective, we love github's data versioning method so much. That's why do the same way to large files

Introduce about iteration data in Rich Anchor Recomendation What is iteration data? According to wikipedia, “Iteration is the act of repeating a process, either to generate an unbounded sequence of outcomes, or with the aim of approaching a desired go…

Can't find big data engineer for your user logging/tracking analytics?
Try our much cheaper method.

Using AWS for collect big data to do user tracking, logging and analytics Have you ever wondered how Google Analytics can build a dashboard which is so awesome. It can report you everything from page views, daily users, real-time active users, time on-site, … That’s also…

Everybody wants big data or distributed computation technology, and we do, too. But before diving, we ask ourselves: Is our data really big?
We've got our selves answer. Check how we're always ready to SERVE 200M USERS with ONE SERVER ONLY.

Is your data really big? In Rich Anchor's backend. We work mainly in algorithms and Machine Learning, which eat server's memory in terrible speed and amount... Millions of users data are really big or not? (Consider it’s 100M users)...

Be careful of building a cross domain recommender. It is poison, which hacked our impressions 20 times in ONE DAY !!!

Cross domain recommendation system in Rich Anchor How we do that, and for what? We've built a version of cross-domain recommendation system for those who want to give recommended items to another system's users. This comes the problems that we nee...

What are you reading? Our recommendations are coming closer, wait for us

Context-based similar documents This post will be the last post in the series: "Text classification - content based recommender system". Recently, we have published two different posts in this series: predict the category of give...

LDA algorithms is Richanchor's new "machine-no-jutsu". We can forget about content label in training data. Now we create any number of categories as we want

Topic modeling with LDA In the previous post about detecting the category of an input text, we have presented to you how to build a simple Naive Bayes classifier for Text classification. In this post, we will show you a m...

We're growing fast now, it's perfect time for talents to join. Guys, we want you!

Join us Our team, our values We are Richanchor team, who develops recommendation system for content provider in scale of South East Asia (about 100M users market). For further information, take a look on o...