I am currently working in a company which deals with automated matching of resumes with job postings. We have really intelligent AI matching engines. We not only deal with simple whole text to whole text matching between resumes and jobs, but also with meticulous information of candidates experience, education, summary, skills set with corresponding job posting. We tag each section then within sugtags then we get more implicit information of location, latitutdes and longitudes of the job location, salary information accurately (everything automated ofcourse!). Finally I would like to mention the name of our company Burning Glass Technologies. Please do let me know if anyone wants to try our products. We train our computers to act like HRs and Recruiters. We hire for you!

Please catch me at http://www.linkedin.com/myprofile?trk=hb_tab_pro

Thanks a lot,

Priyadarshan

Thank you in advance
Daryl Keeley

Good post, and good views on the issues. However, it is important to look at ALL existing matching solutions instead of only those that match against predefined profiles or templates.

Matching should indeed be more than comparing against pre-defined templates. It’s about comparing core data. Text and numerical data. And as this data is always combined in resumes (or jobs), you should be able to search for it in one go; one query containing a combination of structured data, unstructured data, and numerical data. See it as a huge boolean string (however, matching is more than boolean searching, we’ll come to that later) where you can type in things you would like to find, whithout needing to re-query over and over again. When something is not found, matching should find results which are close to the initial criteria looked for. This we call “fuzzy matching”. The results of it should be ranked according the matching level, and according the relevancy. And preferably also say WHY the ranking is like that (to avoid wasting time analysing the results manually).

To make it a little more AI (as this seems to be liked in the initial post , we have also built a huge knowledge base which performs Guided Search. This system looks at the query, looks at the database, and suggests extra words which it found in the database, and which are related to the queried words. It will then indicate how many more, or how much better results it can find if you decided to allow it to search also for those words. Good matching systems allow also to search based on sound. If you don’t know how to spell a name, how to write a word (is Excel with an “x” or “xc” in the middle? with one or two “L”s at the end?). Just type in the word, followed by a special character, and we will find words sounding like “exel”.

So I agree matching is more than comparing against pre-defined templates. It is ok to be able to do so, but that should be an application on top of a matching engine, not the core matching technology itself.

We’ve been around for some time, supplying matching platforms to the HR industry for years, and world wide. It is right to say that it will be difficult to replace humans by technology. But that is also not the objective. The objective is to make the recruitment process as good and as efficient possible, and it is not possible to do that whithout technology. Our systems offer many advantages, even in times of economic downturn. Feel free to look at our web site, or contact me for additional info.

Kind regards,
Bart

wroreHopidork – Ви дуже вітаємо – Дякую за читання. Якщо вам подобається те, що Ви читаєте – будь-ласка, поширенню цього слова. Спасибо!