I have a complete Hadoop platform with HDFS, MR, Hive, PIG, Hbase, etc., Python, R, Java. All data sets have a large size.

The data set A, describing the jobs of people working in a company, is composed of the following fields:

  • Id Person: a unique alphanumeric identifier per person.
  • Start Date: a date format iso entry in the post
  • End Date: iso size release date of the position. If the date is not given, it is the current position

  • Job Title: a text field containing the title and the name of the company. The text is free, non-standardized, French and / or English and can contain typos. Ex: Director Big Data Analytics with Google, Commercial Manager at [missing text] , Manager at googole ...

My question is; how can I create a feature to easily process the name of company of the job (jobtitle)?

Thank you in advance

  • $\begingroup$ You may want to change the question title, since you question title doesn't really match what you want to ask. $\endgroup$
    – fansia
    Feb 10, 2015 at 2:08

1 Answer 1


I think what you want is to extract company names from "Job Title". In natural language process, we call this kind of research as "Name Entity Recognition(NER)". You can try to use Stanford Named Entity Recognizer (NER)[http://nlp.stanford.edu/software/CRF-NER.shtml]. Stanford NER performs very well on English contents and there are lots packages for many programming language:

UIMA: Florian Laws made a Stanford NER UIMA annotator using a modified version of Stanford NER, which is available on his homepage. [Old version.]

Perl: Kieren Diment has written Text-NLP-Stanford-EntityExtract, a Perl module that provides an interface to Stanford NER running as a server.

Ruby: tiendung has written a Ruby Binding for the Stanford POS tagger and Named Entity Recognizer.

Python: Dat Hoang wrote pyner, a Python interface to Stanford NER. [Old version.] NLTK (2.0+) contains an interface to Stanford NER written by Nitin Madnani: documentation (note: set the character encoding or you get ASCII by default!), code, on Github.

F#/C#/.NET: Sergey Tihon has ported Stanford NER to F# (and other .NET languages, such as C#), using IKVM. See also pages on: GitHub and NuGet.

PHP: PHP-Stanford-NLP. Supports POS Tagger, NER, Parser. By Anthony Gentile (agentile).

If you are not satisfied with the performance of Stanford NER, you can also train you own models to extract company names by crawl company names from several popular sites with company names, such as Linkedin/Facebook/Glassdoor...etc

  • $\begingroup$ thank you very much for you reply, it helps me, but is it possible to count the number of people currently working in the same company? $\endgroup$
    – user17241
    Feb 10, 2015 at 9:33
  • $\begingroup$ Yes, after you extract the company names for each people, you can reference the hadoop/spark word count examples to count the number of people in the same company. $\endgroup$
    – fansia
    Feb 10, 2015 at 9:46
  • $\begingroup$ Thank you, Do you think that Hive regex, or Pig regex, can be useful to extract company's names? $\endgroup$
    – user17241
    Feb 10, 2015 at 12:45
  • $\begingroup$ Of course. If you already know the rules to extract all company names, just use the regex to extract the names. $\endgroup$
    – fansia
    Feb 10, 2015 at 13:56
  • $\begingroup$ Ok, i'am trying to use Hive Regex to do this, but I have no idea after this how to count the number of people currently working in the same company. Should I use map reduce algorithm ? $\endgroup$
    – user17241
    Feb 10, 2015 at 14:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.