Emply.ru looking for an investor


A week ago our investor said that for several reasons, his income declined and he can no longer Fund us in full. Since August the team will have to be reduced. To continue the development and promotion we are looking for additional investments.

During the work we made two separate online project, each of which has no analogues on the Russian market:
the

They are unique thanks to the use of our technology for extracting facts from the job boards. Both services differ from existing solutions on the market that allow you to operate not with categories, full text search (text contains "programmer" or "developer" or "developer"), and the categories of the subject area (vacancy programmer).

This approach allowed us to implement a scoring summary — assessment of compliance summary the parameters of the vacancy. This service is for the employer or staffing Agency allows you to save the time of the recruiters at the stage of primary selection summary.
The user is required to specify the parameters of the vacancy (mandatory requirements and evaluation parameters, which will be held scoring) and select which summary to analyze. Summary for analysis can be downloaded independently (for the protection of personal data, we made the program-anonimizator, which leaves the contacts the employer, and emply.ru downloads the already anonymous summary) or to analyze summary from our base, which is filled with spiders (just had 141 636 summary). As a result, the user sees a list of the most suitable for the vacancy candidates. For each specified conformity assessment (percentage).

For text Analytics, we have created a system that we modestly called the "parser". The parser allows you to analyze the texts of job and not just break them down into blocks, as do other parsers summary, and to extract from them the facts (positions, skills, industry, etc.)

To make it all work we have created a knowledge base of the subject area containing the reference of these facts, their possible names (how this object can be called in CV/resume), and a hierarchy of objects, and hundreds of thousands of texts processed manually and resume, which train our ML algorithms.

All this wealth is documented, covered by tests, established the mechanisms for CI, automated deployment. And even a team up to August all in place.
We have a preliminary agreement about the use of scoring with a variety of companies, and we are in talks about integration with the largest Russian ATS (Applicant Tracking System) and the job site.

We welcome any suggestions!
Article based on information from habrahabr.ru

Комментарии

Отправить комментарий

Популярные сообщения из этого блога

Briefly on how to make your Qt geoservice plugin

Изображение
+ =? The next step after GPS module , became its practical application in my project. Maybe someone this post will be more interesting. First, some source data: Required: the Offline mapping service the Integration with QML Features: the the libosmscout — library providing the OSM data in the wrapper with the friendly API the Qt5Location with dev branch summary of the previous 160 series In Qt since time immemorial was a wonderful beam-process QtMobility and he was so wonderful that he lived, he lived, and bent. Yes, of course, he is still here, and even used my N9 and someone's Blackberry, but the fact that Digia will no longer support this API at all. But in Digia are working not fools, they gradually started to pull the pieces QtMobility in Qt5. One such component is a QtLocation. In Qt 5.3, you can see only a piece QtLocation, QtPositioning which refers to Digia. I must say nice stuff, gave her the NMEA from the local socket, and she'
Далее...

Database replication PostgreSQL-based SymmetricDS

In this article I will tell how to configure database replication for PostgreSQL. For the experiments we use a distribution of Linux CentOS 5.3, although this is not essential. we will use a version of PostgreSQL 8.4.7 and SymmetricDS-2.2.2. What is replication? In fact, it is a mechanism to automatically synchronize the contents of databases running on different servers. As a result of replication of these databases contain absolutely identical data. This is necessary for example in order to provide fault tolerance system (in the case of the fall of the first database server, the work enters the second), or to implement load balancing, different customers can serve different server. For replication you need at least two database servers, so prepare two identical servers with a PostgreSQL database on each. The first is the IP address 10.0.2.20, the second — 10.0.2.21, both gateway 10.0.2.2. You can do a virtual machine like VirtualBox to create two virtual servers and run
Далее...

Yandex.Widget + adjustIFrameHeight + MooTools

Изображение
Many people know about this cool thing Yandex.Widget . To make your widget be easy enough to write server-side widget and connect it to Yandex.Widget via iframe . To control the widget Yandex offers APIs in a JS object widget , particularly with the help of it you can change the height of the frame under dynamic content (adjustIFrameHeight), it allows to get rid of the vertical scroll bar. But this API has a significant drawback — it is written using jQuery , which deprives developers of widgets for MooTools to use the Widget API. Below I will show how quite simple it is possible to bypass the Widget API to control the height of the frame using MooTools on the side of the widget. As we know from the contents of the frame it is impossible to access the object frame itself, if they are on different domains. To bypass this limitation, a proxy Widget API uses an iframe using hash (the content after # in the src of the frame) which provides for the exchange of comma
Далее...