JANZZsme! The ultimate semantic
skills and job matching engine.
JANZZsme! is the latest generation of our semantic matching engine for skills and job matching and can also be used for the intelligent application and evaluation of all kinds of occupation (big) data. It can work with both structured and unstructured data, such as:
- Comprehensive labor-market profiles for unemployed individuals and job seekers
- Job offers including job-portal requirements, aggregators or companies’ own job sites
- Profiles from CVs, CV databases or social networks
Via JANZZjobsAPI, the outstanding search and matching functions of JANZZsme! allow for:
- Highly complex queries for extremely precise concept matching (skills and job matching) on a 1 : 1 or 1 : n basis (one-to-one or one-to-many)
- Extensive data mining, for example in the area of occupation data (e.g. ranking of the most frequently requested/offered skills for certain job groups, significant increases in/changes to criteria)
- Considerably improved results in the area of classic full-text searches
- Gap analyses (between job offers, as well as between profiles and job seekers/applicants)
In perfect harmony with JANZZon!, JANZZsme! allows for real, transparent and semantic skills and job matching with a level of precision previously unknown.
JANZZsme! can be integrated and used as a cloud solution in existing web environments and applications in a very simple and cost-effective manner via JANZZjobsAPI. It can be used in connection with, for example, public employment services, job portals and social networks as well as special applications of personnel services providers, recruitment agencies for temporary employees, ATSs and ERP in the area of HCM and staffing and applications in the areas of statistics, training and labor-market data collection.
JANZZsme! – application example: automatic classification of job offers
Over a number of years, the employees of a public employment service of an EU state processed several thousand job offers each month by hand with the objective of assigning them the correct code from the national classification system as well as, for statistical purposes, the ISCO-08 code. The primarily manual process and the processing of hundreds of thousands of offers each year were not only very expensive, but also very labor- and time-intensive. The process was also extremely susceptible to human error, as many job descriptions were not, or only partially, interpreted correctly (including major discrepancies between the allocations of individual employees) and were thus regularly given the incorrect codes.
Improvements made possible by JANZZ.technology
Saving public funds while simultaneously considerably reducing processing times using a procedure that is as automated as possible and guarantees largely error-free allocations within the two classification systems.
Solution and methodology
- Detailed discussions with the project managers allowed for a clear conceptualization of the process as regards the type of data structuring and areas of potential improvement in the context of data entry.
- The public employment service processes all job offers in advance by means of parsing in order to at least make available partially structured text components for further processing. While this pre-parsing is helpful, it was not essential for the processing by JANZZsme!
- The data was processed by JANZZsme!, working in conjunction with JANZZon! that also accessed background and contextual knowledge on the various job titles to ensure that the correct codes could be allocated.
- JANZZsme! then assigned the code for the national job classification system as well as, for statistical purposes, the currently essential international ISCO-08 code to the job offers.
- If it was not possible to assign a code automatically due to unclear or overly general job titles or as a result of repeated contradictory information in the texts, these cases were flagged and separated by JANZZsme! to be checked manually and allocated by experts.
Accomplishment and results for the client
It was possible to substantially and sustainably reduce the processing times and generated costs for a previously very labor-, time- and cost-intensive process with a high susceptibility to errors. The allocations made by JANZZsme! were on average around 42% more accurate, allowing for an improved statistical analysis. Less than 15% of the results had to be flagged for an additional manual check and subsequent allocation due to one of the problems outlined above.