Hire treasure, Algorithm + artificial, create a recommendation-oriented recruitment tool

if there is not enough exposure

"The asymmetry of the recruitment market information and the limitations of the traditional recruiting platform are in fact a contradiction between the limited position and the growing JD." "Fang told me.

For some businesses and their job requirements, it's hard to get enough, matching resumes (and talent) if you can't get enough exposure. On the other hand, even if a candidate does not deliver, it is likely to be a call to a position or company.

Hired Treasure was founded in September 2013, the goal is to become the most efficient recruitment field of third-party referral services, more intelligent and efficient docking enterprises and talent. In the United States, JOBR and hired treasure similar-the former online for several months, they got 2 million dollars angel investment.

"What we're doing is recommending a recruiting product, not a search." ”

I think it's like the idea of general knowledge engine development.

A similar starting point for recruiting products that are recommended for search is to find information more quickly. What's the difference? The scene of the search = you clearly know a request to use the request for information screening. Compared to this, the recommended scenario = You only know the scope of the requirements, you have to read the relevant information to further clarify the requirements. At the same time, the recommendation and personalization are almost natural integration. The same requirements input, different users to obtain the recommended results are not the same.

Hired Treasure Research found: Recruiters in the PO out of certain positions, in fact, it is difficult to decide, well, this job requires you have "more than three years" or "more than two years" related experience--2.5 years of you or not? Or in some companies (such as: BAT) has a working background-in a well-known start-up companies have worked, you think will lose to BAT out of it? When screening conditions are too harsh (and inflexible), companies are likely to miss good candidates. In the case of job seekers, it is often difficult to determine the right industry--electricity, PM, or operation?

The recommended system is to understand the recruitment needs at the same time, according to user behavior constantly revise the recommended results. Referrals are increasingly closer to user preferences, while providing exploratory content.

Exactly how?

Enterprise Login to hire treasure, a simple tick, you can complete the recruitment needs of the input

After the recruitment needs to be resolved, matching, and the algorithm that the matching candidate recommended to the enterprise of a certain recruitment needs

Only 3~5 candidate resumes are recommended at a time to ensure accurate recommendation

When the enterprise receives the recommendation, may choose the direct downloading contact, or sends the job search intention confirmation

When a candidate receives an invitation from an enterprise, it may choose "interested, willing to further contact" or "not interested"

At the same time, the client will record the user's behavior, analysis of user preferences, so that the next recommendation more in line with user needs

The original data from the recruitment of its own IT headhunting team, while the recruitment of treasure also created a "talent partner" independent role-upload idle talent, recommended success to get other resumes download amount. With the help of these 2 ways, there are now more than 100,000 resumes.

In order to cut into the C-terminal faster, the PayPal is expected to launch the December micro-letter version. Job seekers can input part of information anonymously and get job recommendations. When you confirm a job search intention, then enter the full information.

"We consider docking third party voice input interface to simplify the information entry of mobile end users as much as possible." ”

Talking about the technology of employing treasure

Employing the treasure of the algorithm = Large data + artificial optimization-Manual recruitment experience and knowledge system is the basis:

First, it takes a lot of effort to build the "artificial intelligence" of industry recruitment.

For each industry, you will first invite consultants with headhunting or HR experience to discuss and establish the original model and knowledge base. The team also recruited 2 former IT industry headhunters in-house.

Secondly, the design of the algorithm is more in-depth and comprehensive. Better understanding of demand and text resume is the basis of machine algorithm--the matching algorithm of the hired treasure is not only the inclusion relation matching of the pure text, but also extends to the knowledge system, the salary calculation and so on dozens of links.

Again, the goal of the recruitment algorithm is to achieve "mass customization." When the user's behavior is large enough, the algorithm can understand the user's preference more quickly--thus doing the same job demand copy, the recommendation results from different recruiters will vary according to preference differences.

It's not just the same. Internet products, but also a service

As a "service", we hope that no matter what the user is in the scene, can easily get talent or job recommendations.

Access varies in various ways: Web, email, micro-mail.

Imagine a scenario where you receive a job request from a business unit when you work with HR. Next, HR as long as the recruitment request mail forward to the hired treasure, the latter will quickly start matching talent, and then reply to the message will match the results sent to HR. On the other seekers can easily obtain matching job recommendations through micro-letters.

The online version of the hired treasure was launched last December, inviting a small number of users to participate in the experience. Officially released this June, less than half a year has been 1500 enterprise users. Fang told me: the CV recommended by the recruitment of 60% clicks. Interestingly: For job seekers who have not posted resumes on the recommended position, the percentage of "interested" in the job is 40%.