madlib-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dmitry Dorofeev <d...@luxmsbi.com>
Subject Re: Sentiment Analysis
Date Wed, 17 May 2017 15:04:48 GMT
We checked (1) Srivatsan work, but it is almost impossible to reproduce.

(2) and (3) looks interesting, thanks.

----- Original Message -----
From: "Frank McQuillan" <fmcquillan@pivotal.io>
To: user@madlib.incubator.apache.org
Sent: Tuesday, May 16, 2017 7:52:24 PM
Subject: Re: Sentiment Analysis

Here are some links on sentiment analysis using MADlib and/or GPText that I
am aware of:

(1)
Deck on topic from Pivotal data scientist
https://www.slideshare.net/SrivatsanRamanujam/a-pipeline-for-distributed-topic-and-sentiment-analysis-of-tweets-on-pivotal-greenplum-database
Pipeline description starts on slide 18

Github repo corresponding to above
https://github.com/pivotalsoftware/tasa


(2)
Blog on text analytics as a service
https://content.pivotal.io/blog/data-science-how-to-text-analytics-as-a-service

Sentiment classifier using PL/Python on PostgreSQL, Greenplum Database, or
Apache HAWQ, related to blog above
https://github.com/crawles/gpdb_sentiment_analysis_twitter_model


(3)
Blog from zData using Greenplum, GPText and Alpine (which uses MADlib)
http://dewoods.com/blog/alpine-sentiment-analysis


I hope these are useful.  Please let us know how your project progresses.

Frank

On Sat, May 13, 2017 at 1:13 PM, Dmitry Dorofeev <dima@luxmsbi.com> wrote:

> Hi all,
>
> We are a BI developers preparing demo for PGDay'17 Russia. Our demo is
> based on Enron emails dataset and financial data like NYSE stock etc.
> Some data is loaded in Postgres and some data is in GreenPlum, so we can
> use (and already using) GPText and MADlib.
>
> The most exciting thing is sentiment analysis on Enron emails. We want to
> start with email subjects only, which is similar to twits and we found
> several OSS projects which can do that.
>
> Can anybody advise on the best way to do sentiment analysis with GPtext &
> MADlib ? Preferably running inside DB using MADlib?
> Are there any articles, github projects covering GPtext/MADlib sentiment
> analysis you would recommend ?
> What about emails body sentiment analysis, is that easily doable or we
> need to write complex software to do it ?
>
> Thanks
>
> -Dmitry Dorofeev
>

Mime
View raw message