Job IDs are hard to recognize since a job page is usually littered with various types of IDs that resemble a job ID. Jobiak’s learning algorithms can accurately separate and extract the correct job ID from the rest.
Hiring company name can appear anywhere on a job description page. Sometimes part of a large blob of text, sometimes as a image logo on the page or simply implied by the URL. The presence of other company names (like the hosting job board) or company name like entities make it even more difficult to accurately identify the hiring company.
Jobiak’s sophisticated natural language processing and modeling techniques are capable of automatically distinguishing the correct company name from others. This is aided by Jobiak’s proprietary visual/structural parsing technology as well as millions of carefully curated and labelled data.
Job titles are unstructured and can appear anywhere on a job description page often along with other entities like job location, requisition number etc. making it extremely difficult to automatically extract.
Jobiak’s sophisticated natural language processing and modeling techniques utilize 100s of visual, structural and semantic features to recognize and extract job titles with a high degree of accuracy from any unstructured web page. This is aided by Jobiak’s proprietary visual/structural parsing technology as well as millions of carefully curated and labelled data items.
Locations are unstructured, can appear anywhere on the page, often incomplete and along with other entities like job title or in the middle of a large description making it difficult to extract .
Jobiak’s sophisticated natural language processing and modeling techniques are capable of automatically identifying job locations anywhere on the page with a high degree of accuracy as well as canonicalizing it based on contextual information. This is aided by Jobiak’s proprietary visual/structural parsing technology as well as millions of carefully curated and labelled data.
Accurately identifying description is a hard task. Descriptions are made up of large portions of text, often with multiple sections. Accurately identifying text that is part of a job description and identifying the beginning and end of description sections becomes hard, even for human reviewers.
Jobiak employs sophisticated machine learning techniques to identify various sections and topics that are part of the description and accurately classify sections that are part of the description. The technology also uses various algorithms to determine the boundaries of the description so as to accurately extract a description in it’s entirety, no more or no less than what is actually the description.
Jobiak’s algorithms can accurately identify salaries in job descriptions usually written in various formats (ranges), currencies and units (hourly, annually).
Various job types associated with a job are identified whether it is explicitly present in the job page or inferred through context.
Jobiak’s algorithms can accurately detect and distinguish between various kinds of dates like posted dates, validity dates, age etc
Jobiak’s optimization technology is built using sophisticated machine learning algorithms trained using millions of job postings and their online performance over a long period of time. Jobiak has built knowledge structures such as association graphs of titles, skills, descriptions using sophisticated text processing techniques. Convolutional models trained on this data accurately recommend proven job optimizations required to improve online visibility for job listings.