Recognizers Text Versions Save

New release of the Recognizers-Text packages to NPM (https://www.npmjs.com/~recognizers.text). Version 1.3.1

Upgrade dependency on lodash to address vulnerability in older versions.

Also, releasing v1.3.2 of recognizers date time as 1.3.1 was broken.

• [.NET] - Support for NET6.0 Target Framework
• [.NET] - Support for parametric Timeout to RegEx objects
• [EN DateTimeV2] - Two-digit year improvements
• [FR DateTimeV2] - midi expression fix
• [ES DateTimeV2] - Support for the pattern del
• [EN DateTimeV2] - Support for the pattern [minutes] past [hour]
• [IT DateTimeV2] - fix for time expression [minutes] minute alle [hour]
• [FR DAteTimeV2] - Improving the coverage of expressions with "last" in French

• [DateTimeV2] Stack Overflow bug fix in FR
• [NumberRange] support added for PT
• [NumberWithUnit] Mitigations for reported common false positives in EN

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.8.5

Changes:

• [EN DateTime V2] improving support for daterange and cases like "-2020-07-01-"
• [EN DateTime V2] Fixeing the missing resolutions of some duration types
• [EN Number] Added support for "nought" as 0
• [EN DateTimeV2] Improving recognition of informal dates
• [EN Currency] recognizing MUSD as currency
• [JA DateTime V2] DateTimeModel initial support
• [DE DateTime V2] added support for 'en' suffix in German dates containing ordinal numbers as words
• [DE DateTime V2] fixing inconsistent date recognition with ordinal numbers as words
• [PT DateTime V2] fixing the failure to recognize times in the 'das n' format
• [ZH DateTimeV2] adding support for DateTimePeriod patterns like "还剩5 分钟"
• [NL DateTime V2] fixing multiple issues from speech across date/time sub-types

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.8.4

Changes:

• Bug fix for USD Currency in Spanish language;
• Bug fix for DateRange (without year) that would be mispredicted and resolved as a single date;
• Added support for 30-hour clock form (commonly used in Japanese);
• Bug fix for DateTimeV2 where temporal modifiers such as "before" are no longer recognized in multi-mentions like "from 2010 to 2018 or before 2000";
• Bug fix for DateTimeV2 and Currency to fix misc. bugs generated from the speech in ES and PT;
• Bug fix for "ALL" wrongly picked as Currency in English;
• Bug fix for Number in Japanese to recognize the Kanji zero;
• Extending Number and Currency recognition in Japanese to include Kana along with existing Kanji and Arabic numerals;
• Improving the recall on a few Dimension entities in Portuguese;
• Making the interpretation of temporal modifiers "since" and "until" in Chinese consistent with English implementation;
• Bug fix for ATT recognized as Currency in SV;
• Bug fix for DateTimeV2 over mentions in the form "day-of-week, date" not recognized in IT;
• Bug fix for DateTimeV2 to support thw word "ad" in IT;
• Adding support for "as soon as possible" in FR;
• Adding support for mex$ as Currency;
• Supporting "t" as an abbreviation for a ton in NumberWithUnits;
• Bug fix for night terms in German;
• Added "abd dolar" as Currency in Turkish;
• Added the missing support for a few units and abbreviations in ES;

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.8.3

Changes:

• Fixed bugs in EN Datetime including the “since” and “after” modifiers, “a current”, and “greater than” in datetime range and TimexProperty.ToString()
• Fixed Timex for EN ordinals 11-13
• Fixed relative ordinals across languages
• Added Support for angles of rotation in multiple languages
• Added support for informal use of degrees in ES and PT
• Refinements of datetime and NumberWithUnits in PT
• Fixed extraction of phrases like “5 e 45” in PT
• Adding dimensional units support in JP
• Refinement of currency and datetime durations and parser in JP
• Fixed number extraction in datetime and patterns like “is 30 or at least 30” in NumberRange for KO
• Refinement of temperature and Number range in KO
• Refinement of Datetime extraction in parsing in SV
• Refinement of TimeZone extraction and parsing in SV

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.8.2

Changes:

• Fixed errors in number tokenization of digit/character sequence
• Added support for common date format YYYYMMDD
• Added support for unconventional scientific notation
• Fixed ISO Week errors
• Fixed fractional numbers in long form and large number expressed as fraction
• Added support for micrograms as a dimension type
• Fixed issues with emails including capitalization in sequence
• Fixed recognition of relative range in years in EN|ES|PT|CN Datetime
• Fixed fraction parsing from text in forms like “two out of one hundred” in EN Number
• Fixed consistency of units, “%” , “percent” , and “percentage” and appropriate tagging
• Fixed proper handling of transcribed dates for EN Datetime
• Merged Extractions in expressions like “Monday two weeks from now” in EN Datetime
• Fixed consistency in extractions of forms [day abbreviation] [number] (Mon 13th) in EN Datetime
• Fixed Resolution of “last week of this month” in EN Datetime
• Fixed omission for “0” in French numbers
• Fixed ambiguous spelled-out hours in FR Datetime
• Fixed recognition of year when month is spelled out and using “de”
• Fixed date range when starting with a “=” gives stack overflow in FR Datetime
• Fixed failed recognition of time in the n horas format in PT Datetime
• Fixed negative decimals in ZH Number
• Improvements to ordinal and fraction recognition in SV Number
• Refinements of ordinal, percentage and number recognition in KR Number
• Added support for temperature in JP Number

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.8.1

Changes

• Datetime Adding support for holiday-weekend mention patterns
• Fixed 2-digit years in month of year construction in EN/ES
• Fixed recognition of "one morning" in EN
• Fixed recognition of numerical month in ES/PT
• Minor change to year suffix in ES to accept words like del or de
• Fixed tear recognition when month is spelled out in PT
• Fixed times with am/pm modifier to return the specified time when using words like "morgens" in DE
• Supporting Weihnachtsfeiertag in DE datetime
• Support next/last modifiers for dates in DE with nächstem and letztem.
• Fixing inconsistency in recognizing complex entities including fruhmprgens in DE Datetime
• Fixed missed ates surrounded by words such as "avant", "et" "maintnet" in FR datertime
• Adding support for fünfzehn w/ umlaut in DE numbers
• Adding support for units when surrounded by parenthesis/brackets
• Bug fixes for expanded recognition of currency value in FR/ES/PT/US to include [currency name][currency symbol]
• Fixed currency with "con" for decimals in ES
• Support for quoted Text in SV

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.8.0

Changes

• DateTime support in Dutch;
• Bug fix for merged DateTime entities in German where time + weekday range behaviour was inconsistent with English;
• Support for language variations in German for day-of-week and time-of-day DateTime mentions;
• Bug fix for reference year incorrectly assigned to timex in DatePeriod time expressions;
• Bug fix for Time entities post-noon returning inconsistent extra resolution, as if ambiguous, in Chinese;
• Support for "immer" as signal for recurring time (Set) mentions in German;
• Bug fix for weekday + time-range mentions producing an invalid range in French DateTime;
• Bug fix for "hasta"/"até" not properly supported as DataTime range modifier in Spanish and Portuguese;
• Improved false positive filter rules for common cases in English Temperature, PhoneNumber, and DateTime;
• German Holiday recognition coverage improvements;
• Bug fix for relative past DateTime modifiers not always resolved correctly in French and Spanish;
• Bug fix for Duration patterns wrongly normalized/resolved when number is missing in Dutch and English;
• Bug fix for misinterpretation of a relative duration if prefixed by number in Portuguese, French, and Spanish;
• Improved handling of "in" vs "within" in Spanish DateTime ranges;
• Improved support for colloquial Date mentions in Portuguese and English;
• Bug fix for overly aggressive merge of multiple Duration mentions with modifiers in German, Italian, Spanish;
• Bug fix for "Jahr" + year number not properly recognized consistently between German and English;
• Time parser refinements in Japanese.

New release of the Recognizers-Text packages to nuget.org (https://www.nuget.org/profiles/Recognizers.Text). Version 1.7.0

The project has reached over 2.5M package downloads on NuGet/npm/PyPI!

Changes

• Bug fix for incorrect parsing in weekday-date formats in German, French, Portuguese, Spanish, and Italian DateTime;
• Support for NumberRange in French and German languages;
• Improved support for colloquial Date mentions in Portuguese and Spanish;
• Support for early/late modifiers in German DatePeriod;
• Bug fix in assigning subtype for Number with multipliers (e.g., "1.2b");
• Bug fix for "à midi" not always correctly recognized as Time;
• Bug fix for time-of-day entities recognized, but not resolved correctly in Spanish and Portuguese TimePeriod;
• Extended support for Duration terms in English, Portuguese, and Spanish;
• Extended support for expressions indicating the present moment in English DateTime;
• Improved handling of cultures that use multiple Number formats/separators;
• Improved support for merging date/time/timezone terms within brackets in DateTime;
• Improved support for French relative Time mentions;
• Added support for compound Currency entities in Portuguese, Spanish, French, German, and Italian;
• Support for hyphen-connected Unit expressions in German;
• German Holiday recognition coverage improvements;
• QuotedText recognition integration into Sequence recognizers;
• Fixed inconsistency between French and English DateTime entities with article connectors;
• Korean support for Currency and Temperature units (extraction-only);
• Partial Korean support for Dimension units (extraction-only);
• Support Timezone resolution for time-of-day + time patterns (in Preview);
• Timezone resolution fixes for US informal timezone names (Preview);
• Bug fix for over parsing in combining date and time with timezone in English DateTime;
• Improved resolution of 2-digit year mentions in DatePeriod;
• Bug fix in Arabic Number recognizer to handle other culture-specific Unicode number separators;
• Holiday parser refinements in Japanese.