Textkernel Release Notes
Skip to end of metadata
Go to start of metadata
This page contains the latest Textkernel releases. You can select a product that is relevant to you.

Legend


Change Type Description
Added A new feature has been added
Enhanced An existing functionality has been improved
Fixed An issue has been fixed
Deprecated A feature will no longer be supported but has not yet been removed
Removed A deprecated feature has been removed

Adoption effort Level
S Low
M Medium
L High

Configuration?
Wrench icon Indicates a feature that needs to be configured by the Textkernel team. Otherwise the updates are available automatically

Paid?
€ Indicates a paid feature


##########
  • Various SAP-related skills and job titles added to search expansion resources
  • Improved synonyms of Channel Manager

  • Salesforce Integration 2.6: We have fixed a scenario where the admin user could not disable the search and match indexation job using the cancel button.
  • Salesforce Integration 2.6: Enhancement to Search/Match indexation settings, where admin can directly set the limit of indexation jobs to be submitted to the APEX Flex queue. This new control replaces the previous thread limit control, which is now deprecated as we simplify throughput management.
  • Salesforce Integration 2.6: new feature for Recruiter CV import/parsing where optionally a Record type can be defined when creating imported records inside Salesforce

  • CV's that mention a LinkedIn profile now don't show up as having 'social media skills' anymore.
  • City variants like Genova and Genoa now show the same coordinates.
  • Towns in Italy with the same postal codes are now distinguised in Match queries.
  • More city spelling variants have been consolidated: Roma/Rome, Milano/Milan, Napoli/Naples, Torino/Turin, Venezia/Venice.

  • Added a field that indicates whether the job allows remote work or working from home (either part-time, full-time or temporarily)

  • Fixed an issue related to subregion recognition
  • Improved translations of local education levels for multilingual countries
  • Improved recognition of nationalities based on country codes
  • Improved location mapping for French overseas territories
  • Improved recognition of French subregions

  • Improved precision of UK location coordinates when indexing from postal codes
  • Added a warning in case the user is redirected to an external login page, in order to prevent phishing attempts.
  • Redirect to `redirect.jsp` has been fixed (now it supports Search! being deployed to non-root pathname, e.g. home.textkernel.nl/Search-Match3/)
  • The most frequently searched projects/tags are no longer displayed in the user activity report.

  • Skills API lookup defaults to English for missing translations

  • Added Match support for Japanese and Simplified Chinese languages.

  • Salesforce integration 2.5.2: fix issue where sometimes TK indexation process would schedule next job too far in the future.

  • Salesforce integration 2.5: add ability to disable Trigger inside package checking for CV-Attachment changes for cases where customers maintain a Trigger outside of the package
  • Salesforce integration 2.5: Enhancements to batch indexing process, so that it limits the number of indexation batch jobs submitted to the Salesforce APEX Flex queue to better share with other Salesforce processes. Also set a lower default job limit of 20 rather than 100.
  • Salesforce integration 2.5: new feature to support fast indexing of Vacancies, via new parameter on APEX method, including a Platform Event to confirm when the fast index operation is complete.
  • Salesforce integration 2.5: fix on language tag being sent from Salesforce into TK during indexing to improve match results for Salesforce candidates without a CV

  • Improved recognition and splitting of Chilean addresses

  • Salesforce integration 2.4: add support for using Salesforce Attachment object to represent Candidate CVs (previously only supported File object)

  • Salesforce integration 2.3: performance optimisation on queries running in Salesforce to more efficiently support users with millions of Candidate / Vacancy records

  • Salesforce integration 2.2: fix scaling issue affecting customers with several million Candidates where some queries we causing the indexation setting page to load slowly.
  • Salesforce Integration 2.2: change external index field on Textkernel Indexing Status record so that it is always indexed by Salesforce

  • Salesforce integration 2.1.2: add support to data model mapping tool for 3 additional custom objects on Candidate and 5 additional custom objects on Vacancy
  • Salesforce integration 2.1.2: fix error during recruiter CV parsing where status of conversion showed OKOK instead of OK
  • Salesforce integration 2.1.2: enhanced error checking during admin configuration page, so that if the admin navigates away without saving changes they see a pop-up warning telling them to save their changes.

  • Salesforce Integration 2.0.2: Fix scrollbar on side of Textkernel App mapping configuration pane

  • Salesforce Integration 2.0.0: Add documentation for new recruiter CV parsing feature
  • Salesforce Integration 2.0.0: Add widget to import CVs for Recruiter CV Import
  • Salesforce Integration 2.0.0: For Recruiter CV Parsing: allow admin to map parsed data fields to custom environment objects.
  • Salesforce Integration 2.0.0: For Recruiter CV Parsing: allow admin to export/import data mapping to/from a file.
  • Salesforce Integration 2.0.0: For Recruiter CV Parsing: allow recruiters to review parsed data when importing Candidate records
  • Salesforce Integration 2.0.0: For Recruiter CV Parsing: check for existing duplicates when inserting new Candidate records.
  • Salesforce Integration 2.0.0: For Recruiter CV Parsing, store parsed data onto an imported Candidate landing record

  • Salesforce Integration: In indexation monitoring report, show counts of records with a temporary indexing error in addition to showing permanent errors.
  • Salesforce Integration: Admin user can click a button to attempt indexing all Candidate/Job records in SFDC that had a permanent error from previous indexing attempts

  • Fixed location recognition: US Westfield TX
  • Improved German search synonym clusters containing generic terms
  • Improved handling of Sankt/St. variations in German and Austrian Locations

  • Improved search expansions for Italian job titles

  • Various skill extraction improvements for English and French
  • Improved French names of about one hundred skills
  • Added German synonyms for various concepts in the skills taxonomy.
  • Fixed around 100 various skill normalization issues.

  • Improved extraction of LinkedIn profile links in case multiple are mentioned in the document

  • Fixed an issue related to German location parsing

  • Minor improvements to English, French, Dutch and Italian profession classification
  • Improved naming of various project manager jobs in the italian professions taxonomy.
  • Various improvements to French profession normalization
  • Improved classification of various professions in French.
  • Various improvements to the Italian Professions Taxonomy
  • Improved normalization of various terms in the cyber security and energy sectors in the french professions taxonomy
  • Minor improvements to Chinese profession classification
  • Improved classification of media-related professions in Dutch
  • Improved classification of various professions in French, Dutch and Italian.
  • Fixed name and classification of "Technicien Méthodes Process" in the French profession taxonomy.
  • Improved naming of "Lead developer" in all languages.
  • Improved classification of Product Manager in German profession taxonomy
  • Improved normalization of Service employee in Spanish.
  • Better Mapping of 'These' in French Profession Taxonomy
  • Better recognition of plumbing-related jobs in Italian

  • Fixed searching on Github with both a location and keywords
  • Fixed display of result information from Xing

  • Fixed links between translations of jobtitle Analyst/programmer
  • Added more synonyms for social relations manager in French
  • Added various HR software systems to the skills taxonomy

  • Fixed issues with CV upload and editing profile when using IE11 in mobility setups.

  • Updated search and result fields for DPG external source
  • Project names are now displayed correctly in the match indicator

  • Added various healthcare software systems to the skills taxonomy
  • Fixed wrong classification of "Betreuung Firmenkunden" in profession taxonomy

  • categories parameter for skill autocomplete is now optional

  • Added various healthcare software systems to the skills taxonomy
  • Add "COVID-19 Testing" skill to the skills taxonomy.
  • Added 13 skills and 300 synonyms to the French skills taxonomy

  • Removed the old website pages (about Jobfeed, news, contact)
    This is now all handled through the brand new www.jobfeed.com page

  • 300 English job titles added to autocomplete and synonyms

  • Improved deduplication rules for all Jobfeed countries

  • When searching for active jobs, expired postings belonging to that job would not be searched in.
    This has been changed, improving your chance of finding the right job lead

  • Improve University Name splitting
  • Linkedin PDF: improved candidate location extraction
  • Linkedin PDF: Parsing improvements for fields that are line wrapped
  • LinkedIn PDF and Indeed CVs: Improved address recognition

  • Location parsing in France vacancies: no longer report subregions as vacancy location

  • Added accounting standards to French skill taxonomy
  • Fixed spelling issue "Espagnole"
  • Evaluation of French SNT feedback. Includes the creation of 13 new professional skill concepts and the renaming (French-only) of 3 existing skill concepts.

  • Autocomplete endpoint added

  • Improve address fields for locations from Ireland

  • Prepared Jobfeed to allow customers to match from within Jobfeed to their internal candidate database with a new field called seniority
  • Improved classification of direct employer companies
  • Customers can now try out the new lead generation screens in Jobfeed
    When the remaining screens are ready the old screens will disappear

  • Evaluation of French SNT feedback. Includes the creation of 13 new professional skill concepts and the renaming (French-only) of 3 existing skill concepts.
  • Improved names of 50 French skills
  • Added hairdresser-related skills in French
  • Fixed issues with extraction of 'results analysis' skill
  • Fixed wrong translation of Scaled Agile Framework in German
  • Fixed erroneous classification of 'Afghaans' in Dutch documents
  • Added 70 skills to the English skills taxonomy, mostly media-related

  • Fixed an issue where, in some cases our Bullhorn native integration was unable to import some of the custom text field data from the customer environment to the Textkernel Search & Match index.
  • Removing Lead Entity support from Bullhorn native integration indexation. The Lead is an optional entity, which only some Bullhorn customers use, and it is not a useful concept for our Candidate/Job sourcing and matching use case, so we are no longer indexing it.

  • Fixed an issue with the recognition of 'Lisboa'
  • Resolved an issue with location recognition in Austria
  • Improved recognition of 'QA' related jobs and skills
  • Improved recognition of French accountancy-related job titles
  • Improved recognition of (sub)regions in india and various other countries

  • Moved 'business consultant' to a more appropriate group in the professions taxonomy
  • Fixed wrong classification of photovoltaics-related professions
  • Gender neutralisation of two English professions (Sportsman and Handyman)
  • Several improvements of French and Dutch job classification
  • Improved classification of service assistants in French
  • Gender-neutralization of all profession names in profession taxonomy
  • Improved descriptions of a selection of managerial professions in French
  • Improved descriptions of translator and elementary school teacher professions in Dutch profession taxonomy
  • Improved normalization of IT jobs in Italian
  • Better classification of bank staff nursing professions
  • Improved classification of payroll and accounting jobs
  • Improved recognition of "Erzieher" in German profession taxonomy
  • Improved French translation of Channel Manager
  • Improved classification of design-related jobs
  • Improved classification of personal assistant jobs
  • Added three new professions to the profession taxonomy, and improved the French names of a few existing ones
  • Various improvements to Italian job classification
  • Improved profession classification for travel agents in Dutch

  • Based on client feedback, instances were added to the Profession Normalization Table to improve (mainly French) profession mapping.
  • Improved classification of wood-processing professions
  • 18 new professions added to the profession taxonomy

  • Fixed erroneous classification of 'Afghaans' in Dutch documents
  • Improved recognition of the country Austria in German documents
  • Improved recognition of (sub)regions in india and various other countries

  • New field that captures the experience level (seniority) of a candidate or required by a vacancy.

    Can be used both for searching and matching. When enabled in Match queries, it leads to improvements in matching quality

  • Added 1100 French terms to the skill taxonomy
  • Added 1100 Dutch terms to the skill taxonomy
  • Added 70 synonyms to the French skills taxonomy

  • New option to disable automatic word wrapping for input formats that support soft wrap (e.g. Microsoft Word). PDF files do not support soft wrap. When enabled, longer lines are no longer wrapped at 80 character limit. Useful for customers that build an UI to present the parsing results

  • Added 950 keywords in 6 Dutch, German, English, French, Spanish and Italian

  • Improved detection if a company is a staffing agency or direct employer

  • 800 synonyms added to German skills taxonomy
  • Added 1100 French terms to the skill taxonomy
  • Added 1100 Dutch terms to the skill taxonomy

  • Added cloud-related skill synonyms
  • Improved synonym recognition for operation managers

  • ISCO profession descriptions are now available in German

  • Previously, in some cases, if an existing Bullhorn username was changed inside Bullhorn, that user could no longer be synced into Textkernel portal to allow them to access Textkernel. After this change, Textkernel will allow the username to change in Bullhorn.

  • Added a few IT skills to the German and English skill taxonomies
  • Fix casing of articles in French skill descriptions (e.g. "Science de L'Organisation" > "Science de l'Organisation").

  • Further improve staffing/recruitment agency detection based on company name for Jobfeed DE, NL, FR, CA, ES, IT

  • Improved CV2job Matching by better understanding of the job titles from the CV (e.g. "Java developer - Machine Learning" is made of the "Java developer" job title and the "Machine Learning" computer skill)

  • Bullhorn accounts can now be configured with an additional Bullhorn action menu item that will 'Rank Applicants' when opening Textkernel Match results from a Bullhorn JobOrder. In this case, the displayed matching Candidates are filtered to show only those with JobSubmissions and/or WebResponses against the Bullhorn JobOrder, so that recruiters can quickly see the best applicants for a role to decide which ones to progress.

  • Added salary graph to Germany, Belgium, France, Spain, Italy and the Netherlands
    Added a geographical map for Austria and Spain
    Added salary search option for Italy

  • Compatibility for Taleo 20D.0.8

  • Improved posting deduplication logic for Canada and Spain

  • Added 24.000 websites to Jobfeed Austria.
    Jobfeed will check every day if these sites contain new jobs
  • Added 40,000 websites to Jobfeed Canada.
    Jobfeed will check every day if these sites contain new jobs
  • Added 230,000 websites to Jobfeed UK.
    Jobfeed will check every day if these sites contain new jobs
  • Added 1.2 million websites to Jobfeed Germany.
    Jobfeed will check every day if these sites contain new jobs
  • Added more Pharma companies to Jobfeed Germany, Spain, UK and US
  • Added 60.000 websites to Jobfeed France.
    Jobfeed will check every day if these sites contain new jobs
  • Added any missing stock exchange companies to the 10 Jobfeed countries
  • Validated Jobfeed France has the 300 largest municipalities correctly spidered and stored
  • Improved distinction between qualitative jobboards, aggregrators and direct employers in Jobfeed France
  • Added thousands of new websites to Jobfeed UK and Jobfeed US
  • Jobfeed is now technically ready to investigate 1 million NEW websites every week (on top of the millions of websites it was already reading)

  • Entering a zip code in the city field when explicitly mentioning it's a city name combined with a search radius would not show any results.
    Jobfeed is now able to resolve that situation
  • Some virus scanners from customers would test every job link in the excel email alert
    With potentially thousands of jobs in a single spreadsheet this caused some errors
  • The API endpoint did not correctly return the labels for multi-valued fields
  • Jobfeed Belgium: Added education level
    Jobfeed Italy: Added ISCO, Education level, Contract type and Employment type
    Jobfeed Spain: Added Contract type
  • Users who purchased the skills feature can now search for jobs requesting certain skills, setup email alerts based on skills and filter on skills in the Insights page
  • Jobfeed Italy: Added salary, education level, contract type and employment type to the API
    Jobfeed UK and Belgium: Added education level to the API
    Jobfeed Spain: Added contract type to the API
  • It's now possible to search for a city with a smaller radius than 10 kilometers/miles in Jobfeed Canada and Jobfeed Spain
  • Jobfeed information is now available in a graphical form which can be embedded in ATS systems with a single line of code
  • New online help feature available in various Jobfeed countries for users that speak French, English and Dutch
  • Custom job categories can now also be made in Jobfeed Italy
  • Added the following information to Jobfeed
    Jobfeed Germany: Duplicate on Jobsite, added/cleaned up posting language options, salary slider
    Jobfeed France: Duplicate on Jobsite, added/cleaned up posting language options, salary slider
    Jobfeed Netherlands: Salary slider
    Jobfeed Belgium: Salary slider
    Jobfeed Austria: Added/cleaned up posting language options, salary slider
    Jobfeed Italy: Source type, duplicate on Jobsite, added/cleaned up posting language options
    Jobfeed Spain: Added/cleaned up posting language options, salary slider, employment type
    Jobfeed Canada: Contract type, added/cleaned up posting language options
  • Added normalized professions to the Jobfeed Italy search screen including autocomplete. Professions have also been added to Jobfeed Analytics
  • Adding Insights to Jobfeed Italy showing number of jobs, posting duration, trend, Job Board distribution, top professions, skills, top employers, top agencies and regional distribution
  • Jobfeed vacancies can now be exported as leads to the Bullhorn ATS
  • Advertiser type (staffing agency or direct employer) indicator added to Jobfeed Italy
  • Jobfeed Italy screens all translated to Italian
  • Jobfeed now shows all requested skills in the jobs as of 2017
    They can be found in Insights and Analytics (if purchased)
  • Switched to a new database release which increased the speed and stability of Jobfeed.
  • Improved deduplication of jobs
  • For Jobfeed UK, Belgium, Austria and Spain the old URL's will go away as of the 4th of January 2021
    The new domains will be:
    uk.jobfeed.com
    be.jobfeed.com
    at.jobfeed.com
    es.jobfeed.com
    Anyone still using the old URL will be redirected to the new URL. Note: This redirect will not work for API calls

  • Added two clusters of German automotive skills

  • Error severity levels have been revised and updated for the Extract SOAP API, including advice on retrying strategies for better error handling. Some error IDs have been changed. For more details, refer to the Sourcebox API Reference documentation.

  • Fixed a deduplication issue in Canada and Spain that resulted in some postings not being deduplicated when street or neighborhood information was present in one of the postings

  • Improved parsing speed and reduced timeouts for certain documents when highlighting of extracted values in the HTML rendering is enabled

  • Postings in which only a country name was mentioned as working location, were geocoded in the center of the country. New postings for such cases will no longer receive any geocoordinates.

  • Improved extraction of names from LinkedIn profiles
  • Improved detection of Indeed profiles so that those profiles are parsed with high accuracy

  • Fixed a login issue occurring in some integrations when using Google Chrome 84.0
  • The first letter of an a suggested query term was sometimes duplicated.

  • Improved keyword extraction for English Match queries

  • Improved country detection for Australia
  • Improved splitting of names for French
  • The highest education level is no longer derived from unfinished education items

  • Improved classification of Freelance and Apprenticeship contract types for French
  • Added language guessing when no full text or document language is provided via the Input API

  • Various improvements to extraction of LinkedIn profiles
  • Improved extraction of name from LinkedIn and Indeed profiles

  • Enriched match queries for Polish with the Textkernel profession classification, resulting in higher relevance of the match results
  • Automatic cleaning of noise from job titles in the Match query is now active for Italian, Spanish, Simplified Chinese, Portuguese, Japanese and Polish. Cleaning was already active for English, German, French and Dutch.

  • Internal enhancements to the Talent-soft data indexation component to increase stability and fault tolerance when the Talent-soft Vacancy data is partially corrupt.

  • New feature for the Talentsoft Marketplace Match integration for the user can set a query term / tag as either Mandatory (Must-have) or Optional (Nice-to-have).

  • Bullhorn users were not able to import candidates from external sources.

  • When using Textkernel's Bullhorn Native integration, and importing from an external searcher to create a new record in Bullhorn, the Candidate Inbox document should be associated to the user's Bullhorn Native User ID, instead of a shared user (i.e. 'export'). This means that documents will now be listed in 'My Documents' in the Candidate Inbox.

  • Improved extraction of posting dates for Dutch

  • When using Textkernel's Bullhorn Native integration, and importing from an external searcher to create a new record in Bullhorn, the Candidate Inbox document should be associated to the user's Bullhorn Native User ID, instead of a shared user (i.e. 'export'). This means that documents will now be listed in 'My Documents' in the Candidate Inbox.
  • For users with admin rights, it is now possible to export users information in CSV format, including the status of the user and usage dates, among others. This improves the user experience, who until now relied on copy-pasting information from the screen into a spreadsheet for their reporting needs.
    Note: the new option can be found under the 'Users' button on the top left side of the screen. This button used to be labeled 'Add Users'.

  • Updated Polish, Slovak and Hungarian UI translations for sorting options
  • Added UI translations for all languages for the hide/show toggle of the query panel

  • Added support for Monster Canada as external source
  • Canadian postal codes in certain formats were not recognized
  • Improved ordering of autocompletions in case of multiple exact prefix matches on the user's input. The order is now alphabetical instead of reverse alphabetical.

  • Improved classification of contract types for French

  • Facet counts for nested fields returned by the API now take into account filters that are set on the first level of the nested field. For example, it can be used to filter by job ID and get the number of candidates per pipeline status for that job ID, or to filter by language skill and get the number of candidates per language level for that language.
  • Added DPG as external source for The Netherlands, which replaces Nationale Vacaturebank and Intermediair. Importing candidates is not yet supported.
  • Changes to a nested query field in the UI (e.g. location + radius or language + level) were not applied to other search tabs
  • In some cases, the name of a person on LinkedIn (using Google API) was not returned

  • Improved geocoding for France in case a postal code or only a department is mentioned

  • Enriched match queries for Simplified Chinese and Japanese with the Textkernel profession classification, resulting in higher relevance of the match results
  • Improved match relevance in case years of experience with no upper bound was requested in a job description

  • Improved geocoding for France in case a postal code or only a department is mentioned

  • The profession code & class have been removed from Match queries, keeping only the profession group. This makes the query more user-friendly. The profession code & class were not contributing to the relevance of the results.
  • Added Text Fingerprint to Match queries. The fingerprint contains a large amount of keywords from the original profile (person or job), and boosts results containing salient keywords over common keywords. This way, it captures important nuances and greatly improves the ranking of the match results. The fingerprint is not displayed to users in the Search UI, but is active in the background.

  • Querying with a field value containing the '$' character resulted in an error
  • In certain browser versions, the user's UI session was terminated after a validation error on the "save project" dialog.

  • In the Bullhorn-native integration, de-duplication was originally designed for single-tenant environments. It has been extended to work with multi-tenant orgs so that Bullhorn BusinessUsers will only see possible duplicates according to their Departments. This does not require configuration.

  • Improved logic for computing annual salary for Austria

  • Added support for Taiwanese Address parsing

  • High accuracy parsing of kariyer.net profiles (Turkey)
  • Parsing of Turkish, Slovenian and Croatian CVs. Documents in these languages can now also be used in Search. Semantic Search is available for Slovenian and Croatian.
  • Improved country detection for French addresses containing German city names as street name
  • Reduced the amount of documents going through OCR when no actual OCR was needed. As a result, parsing is faster and more accurate for these documents.

  • Added support for Whatfix integration in the UI
  • Search & Match API documentation is now available via developer.textkernel.com
  • Initializing the UI with a 'query state' on the URL is no longer supported due to browser limitations on the length of the URL. Use IDs to load saved searches via URL.
  • Improved UI load time by optimizing loading of saved projects/tags. Added a new optimized projects API version (v5).

  • Almost no results were returned when searching in Stack Overflow (via Google API)

  • The keywords in Match queries are now more relevant. Unnecessary keywords have been removed from the query, which makes it easier to digest for users.

  • Improved extraction of phone numbers and language skills from LinkedIn PDFs

  • Improved classification of language proficiency, in particular level B2.

  • Improved readability of external profiles for certain sources (in particular VDAB, CV-Library and Resume-Library)
  • Re-added Dice (US) external source due to Dice platform changes

  • Improved classification of language proficiency levels C1 and C2.

  • Improved Matching when a minimum or maximum number of years of experience is required as mentioned in a job description

  • When unchecking certain related terms in the UI, other related terms were unintentionally unchecked as well

  • Improved classification of contract types for French

  • Improved relevance of Match results in case the job title contains "assistant to..." or similar

  • Improved name splitting for German and Dutch when titles like MSc, MA, BSc, BA are part of the name

  • When importing a candidate from an external searcher into Bullhorn, the owner (or recruiter) will be set to the user doing the import, instead of the Textkernel API user.

  • No labels