Skip to content

Releases: microsoft/Recognizers-Text

Release of Recognizers-Text Nuget packages v1.6.0

20 May 04:26
c70ff6f
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.6.0

Changes

  • Extended support for date + time forms in Portuguese DateTime;
  • Extended support for unambiguous date formats in DateTime;
  • Add support for emoji skin tone modifiers across cultures in Choice recognizer;
  • Add support to handle common mispelt ordinals in English Date;
  • Additional support for non-standard speed units in English NumberWithUnit;
  • Support for bitcoin and its Unicode symbol as Currency;
  • Splitting clustered units into their separate Unit entries in English;
  • Bug fix resolving time-of-day modifiers in DateTime in French and Spanish;
  • Support for million/billion/trillion Number abbreviations in English Number;
  • Bug fix for false positive hours incorrectly extracted from float number in DateTime;
  • Improved performance in recognizing long Number forms in Japanese;
  • Bug fix in resolution for "anoche" in Spanish DateTime;
  • Improved support for merged timex of duration/datetimerange (e.g., "PT1H30M") processing in TimexLib;
  • Bug fix to leap year resolution and output format for invalid dates like "2/29/2019" in DateTime;
  • Bug fix in numbered week resolution (e.g., 2021-W02) in TimexLib;
  • Bug fix in Timex parsing across different cultures in TimexLib resolver.

Release of Recognizers-Text Nuget packages v1.5.0

06 Jan 08:17
b531199
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.5.0

The project has reached over 2M package downloads on NuGet/npm/PyPI!

Changes

  • Expanded coverage for DateTime mention formats across sub-types in Spanish;
  • Bug fix in Spanish NumberRange when multiple non-mergeable sub-ranges are present in input;
  • Bug fix in inconsitent normalization of Spanish Date ranges;
  • Bug fix in resolution for years spelled as words in English DateTime;
  • Bug fix in normalization and resolution of relative year mentions in German DateTime;
  • Bug fix for regression in Chinese DateTime handling "western formats";
  • Improvements support for relative DateTime expressions like "el año anterior" in Spanish;
  • Support for relative Holiday calculations in DateTime;
  • Revised support for Date ranges in Spanish (cleanup and new forms);
  • Revised support for large numbers and informal forms in Chinese;
  • Support for composite durations in Chinese DateTime;
  • Fix for overly aggressive entity merging in Spanish DateTime;
  • Improved handling of fractions and percentages in Chinese;
  • Improved handling of potentially ambiguous terms in Chinese Number and Dimension;
  • Support for Unicode vulgar fractions in .NET across western languages;
  • Bug fix for entity boundary issue in English recurrent dates (Set);
  • Bug fix in support for superscript 'a' and 'o' in Spanish Ordinal;
  • Support for informal abbreviations and prefixes in English Age;
  • Currency support in Swedish;
  • Performance improvements in Swedish Number;
  • Support for Chinese dynasties as Date ranges/periods;
  • Bug fix in Chinese DateTime support for year ranges and decades;
  • Expanded coverage for Date expressions in French (ongoing).

Release of Recognizers-Text Nuget packages v1.4.2

11 Nov 02:11
9715828
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.4.2

Changes

  • Support for Age, Temperature, and Dimension units in Swedish;
  • Improvements to handling informal DateTime ranges and Duration in French;
  • Improved recognition of relative ranges and periods in Spanish DateTime;
  • Support for DateTime entity mentions in the form "[n] [date-unit] from [datetime]";
  • Bug fix in handling Currency ISO codes before monetary amounts;
  • Improved parsing of fractions (including textual and unicode fractions) in Number recognizer;
  • Exposed sub-type information in extracted Dimension unit entities (weight, speed, etc.);
  • Bug fix handling recurring times (Set) referencing weekends in English;
  • Bug fix parsing entities in sentences with multiple DateTime ranges mentions in English;
  • Implemented handling of fractional Number term in German;
  • Improved support for fractional Number in English;
  • Implemented merging of compound Unit entities;
  • Improved handling of "start/end of" in DateTime ranges in English and Spanish;
  • Support for terms like "work day" and "work week" in German DateTime;
  • Bug fix for null resolutions during processing of certain Date ranges in Spanish;
  • Bug fixes in handling "quarter", "around", "now", "weekend", and abbreviated months in Spanish DateTime;
  • Bug fix handling whitespace as Date separator in French;
  • Refined coverage for modifiers (early/earlier/late/later, next/past) in Spanish DateTime;
  • Bug fix handling lists of years in DateTime;
  • Bug fix in disambiguating "morning"/"tomorrow" in German DateTime;
  • Bug fix for Set and Time entities being incorrectly mixed during recognition in German;
  • Bug fix for Holiday + Time not properly extracted and parsed in German;
  • Expanded coverage for Time range and time-of-day expressions in German;
  • Bug fix in parsing complex DateTime ranges in Spanish and English (e.g., "since A and not after B");
  • Improved handling of "more than" Number ranges in Chinese;
  • Support for "万" as multiplier in Chinese Number;
  • Bug fix handling suffix for "half" in Chinese Number.

Release of Recognizers-Text Nuget packages v1.4.1

27 Aug 01:44
a0d465a
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.4.1

Changes

  • The recognizers are now .NET Core 3.1 compatible;
  • Improved support for relative Date Range in German;
  • Improved extraction coverage in Arabic Numbers;
  • Fix compound disjoint Number Range extraction bug in English and Spanish;
  • Hindi DateTime improvements with focus on ranges/periods;
  • Support for "fiscal year" in Spanish DateTime;
  • Resolution improvements for year Date Range in Spanish;
  • Better support for relative Holiday mentions;
  • Improved Holiday support for Dutch (coverage and fixes);
  • Support for part-of Date Range mentions (e.g., "by the end of this month") in English;
  • Support for "to/till date" as Date Range in English;
  • Bug fix in Chinese Currency parsing;
  • Extended Dutch Currency support;
  • Improved support for Numbers like "dozen" in French;
  • Bug fix in initialization of URL recognizer when no culture is specified;
  • Add strict/relaxed match and validation to E-mail recognition;
  • Bug fix were elided Numbers were wrongly extracted in Italian.

Release of Recognizers-Text Nuget packages v1.4.0

16 Jun 12:53
3e77064
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.4.0

In June/2020 the project has reached 1.25M package downloads across platforms!

Changes

  • Improved recognition of Japanese Number (Cardinal), Ordinal, and Percentage;
  • Improved recognition of Dutch Number (Cardinal) and Ordinal;
  • Support for NumberRange in Dutch and Hindi;
  • Multiple improvements to Spanish NumberRange;
  • Support for the Indian numbering system in English Number;
  • Improved handling of relative modifiers in German DateTime;
  • Recognition of dialectal Time expressions in German;
  • Support for informal Time mentions in Portuguese;
  • Multiple refinements in DateTimeRange resolution using boundary context;
  • Support for anchored day of week in parsing relative week entities;
  • Improved recognition of Hindi Time, TimeRange, and Duration;
  • Improved handling of variants and gender in French Number;
  • Bug fix for weekday timexes in Portuguese DateTime;
  • Bug fix in German DateTime to properly handle merging weekday + time_of_day;
  • Improved parsing of fully written Dateentities in Spanish and Portuguese;
  • Bug fix in Duration in Spanish and Portuguese;
  • Improvements to false positive extractions of *Ranges from phone numbers;
  • Improvements to false positive extractions of Unit and Time in Chinese and Japanese;
  • Fix in TimexRangeResolver to handle times with date constraints;
  • Improved handling of UTC reference times in TimexRangeResolver;
  • Improvements to handle Number false positives in Chinese;
  • Bug fix for parsing month + two-digit year in Portuguese, Spanish, and French DateTime;
  • Support for part of day in French DateTime;
  • Extensions to TimeZone handling of European forms and extra non-standard timezone names (in Preview);
  • Improved handling of modifiers like "end/beginning/middle" in year ranges;
  • Extended support for multipliers in handling Number/NumberRanges (e.g., "5k-20k", "20MM");
  • Improved handling of month and day of week abbreviations in French Date/DateRange;
  • Improved recognition of Phonenumber corner cases;
  • Support for approximate DateTime in Spanish;
  • Support for "night" and "weekend" ranges in English DateRange;
  • Extension in Hindi Holiday to recognize additional lunar holidays;
  • Holiday fixes/extensions for Easter, Worker's/May day, Juneteenth, etc. resolution;
  • Bug fix in Portuguese parser for relative past Time;
  • Improved support for PRESENT_REF entities in German DateTime;
  • Fix to reduce false positives in French Unit;
  • Bug fix in French handling of "summer" in DateRange;
  • Support for multipliers/dividers in English Set;
  • Support for "weekdays" in English DateRange and Set;
  • Support for "business hours" in Spanish TimeRange;
  • Improvements to Hindi Set support;
  • Support for NumberRange in Japanese - Extraction-only;
  • Support for Number (Cardinal), Ordinal, Percentage, and NumberRange in Arabic - Extraction-only;
  • Support for Ordinal, Percentage, and NumberRange in Korean - Extraction-only.

Release of Recognizers-Text Nuget packages v1.3.2

12 Mar 06:26
7ea334b
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.3.2

Changes

  • Support for DateTime recognizers in Hindi;
  • Multiple refinements in Spanish and Portuguese DateTimeRange resolution;
  • Extension in Chinese to handle extra relative DateTimeRange scenarios and "≤" and "≥" unicode chars;
  • Bug fix in French causing false positives in Date extraction;
  • Bug fix in French causing incorrect span calculation for some extracted DateRange entities;
  • Text library now offers methods to convert span indexing between 'char-based' and 'text-element-based';
  • Extension in English DateTime to account for new non-standard written date forms.

Release of Recognizers-Text NPM packages v1.3.0

25 Feb 23:56
ce44290
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to NPM (https://www.npmjs.com/~recognizers.text). Version 1.3.0

Major update to bring it closer to parity with .NET/NuGet.

Release of Recognizers-Text Nuget packages v1.3.1

25 Feb 20:09
ce44290
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text).

Retracted due to incorrect nuget package dependency reference, please use packages version 1.3.2.

Release of Recognizers-Text Nuget packages v1.3.0

17 Dec 11:29
51c3ce7
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.3.0

Changes

  • Support for Cardinal, Ordinal, Percent recognizers in Hindi;
  • Support for Age, Temperature, Dimension, Currency in Hindi;
  • Support for Choice recognizers in Turkish and Hindi;
  • Support for NumberRange recognizer in Turkish;
  • Multiple refinements to Turkish DateTime support across sub-types;
  • Improvements to German DateTime support for colloquial scenarios;
  • Bug fix in Spanish DateRange support for cases that cross year boundaries;
  • Bug fix in DateTime to avoid cases where time expression loses generality;
  • Extension to English recurring Set for scenarios like "every other ", "quarterly";
  • Extension to English to handle informal TimeRange and "week" mentions;
  • Extensions and bug fix in Ordinal recognition in Chinese;
  • Refinements to Currency parsing;
  • Improvement to time expression library for week-in-month scenarios;
  • Bug fix in DatePeriod extraction;
  • Multiple performance (latency) improvements;
  • Removal of support for .NET 4.5 and 4.5.2.

Release of Recognizers-Text Nuget packages v1.2.9

11 Nov 07:54
a158808
Compare
Choose a tag to compare

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.2.9

Changes

  • Support for DateTime recognition in Turkish;
  • Support for Swedish Ordinal and Percent recognition;
  • Refinements to Spanish Date and DateRange resolution;
  • Bug fix in DatePeriod extraction.