Class SubsegmentSearchSettings

Used to specify parameters to apply when subsegment matching

Inheritance
Object
SearchSettings
SubsegmentSearchSettings
Inherited Members
Object.ToString()
Object.Equals(Object)
Object.Equals(Object, Object)
Object.ReferenceEquals(Object, Object)
Object.GetHashCode()
Object.GetType()
Object.MemberwiseClone()
Namespace: Sdl.LanguagePlatform.TranslationMemory
Assembly: Sdl.LanguagePlatform.TranslationMemory.dll
Syntax
public class SubsegmentSearchSettings : SearchSettings
Remarks

TM-TDB matching (aka 'baseline' subsegment recall) process fragments of the query segment, and for each fragment, searches for a complete source TU segment matching it. If found, the complete target TU segment is proposed as a translation of the query fragment. DTA matching (aka 'full' subsegment recall) process fragments of the query segment, and for each fragment, searches for a source TU segment that also contains that fragment. If found, it uses additional processing (such as fine-grained alignment information) to identify a fragment of the target TU segment to propose as a translation of the query fragment. Each type of matching can be subject to a minimum length of matching fragment. Because some tokens are essentially valueless for recall (e.g. whitespace, tag tokens), the length is calculated by counting the number of 'feature tokens' in the fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.GetFeatures(System.Collections.Generic.List{Sdl.LanguagePlatform.Core.Tokenization.Token},System.Globalization.CultureInfo,System.Collections.Generic.List{System.Int16}). In addition, some match fragments may be of no interest for recall if they consist only of tokens with little translational significance, such as stop words. So, each type of matching can be further subject to a minimum number of significant feature tokens in a matching fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.IsSignificantForRecall(Sdl.LanguagePlatform.Core.Tokenization.Token)).

Constructors

SubsegmentSearchSettings()

Used to specify parameters to apply when subsegment matching

Declaration
public SubsegmentSearchSettings()
Remarks

TM-TDB matching (aka 'baseline' subsegment recall) process fragments of the query segment, and for each fragment, searches for a complete source TU segment matching it. If found, the complete target TU segment is proposed as a translation of the query fragment. DTA matching (aka 'full' subsegment recall) process fragments of the query segment, and for each fragment, searches for a source TU segment that also contains that fragment. If found, it uses additional processing (such as fine-grained alignment information) to identify a fragment of the target TU segment to propose as a translation of the query fragment. Each type of matching can be subject to a minimum length of matching fragment. Because some tokens are essentially valueless for recall (e.g. whitespace, tag tokens), the length is calculated by counting the number of 'feature tokens' in the fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.GetFeatures(System.Collections.Generic.List{Sdl.LanguagePlatform.Core.Tokenization.Token},System.Globalization.CultureInfo,System.Collections.Generic.List{System.Int16}). In addition, some match fragments may be of no interest for recall if they consist only of tokens with little translational significance, such as stop words. So, each type of matching can be further subject to a minimum number of significant feature tokens in a matching fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.IsSignificantForRecall(Sdl.LanguagePlatform.Core.Tokenization.Token)).

Fields

MinFeatures

The minimum length of a DTA match fragment (see SubsegmentSearchSettings remarks)

Declaration
public int MinFeatures
Field Value
Type Description
Int32
Remarks

TM-TDB matching (aka 'baseline' subsegment recall) process fragments of the query segment, and for each fragment, searches for a complete source TU segment matching it. If found, the complete target TU segment is proposed as a translation of the query fragment. DTA matching (aka 'full' subsegment recall) process fragments of the query segment, and for each fragment, searches for a source TU segment that also contains that fragment. If found, it uses additional processing (such as fine-grained alignment information) to identify a fragment of the target TU segment to propose as a translation of the query fragment. Each type of matching can be subject to a minimum length of matching fragment. Because some tokens are essentially valueless for recall (e.g. whitespace, tag tokens), the length is calculated by counting the number of 'feature tokens' in the fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.GetFeatures(System.Collections.Generic.List{Sdl.LanguagePlatform.Core.Tokenization.Token},System.Globalization.CultureInfo,System.Collections.Generic.List{System.Int16}). In addition, some match fragments may be of no interest for recall if they consist only of tokens with little translational significance, such as stop words. So, each type of matching can be further subject to a minimum number of significant feature tokens in a matching fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.IsSignificantForRecall(Sdl.LanguagePlatform.Core.Tokenization.Token)).

MinSignificantFeatures

The minimum number of significant tokens in a DTA match fragment (see SubsegmentSearchSettings remarks)

Declaration
public int MinSignificantFeatures
Field Value
Type Description
Int32
Remarks

TM-TDB matching (aka 'baseline' subsegment recall) process fragments of the query segment, and for each fragment, searches for a complete source TU segment matching it. If found, the complete target TU segment is proposed as a translation of the query fragment. DTA matching (aka 'full' subsegment recall) process fragments of the query segment, and for each fragment, searches for a source TU segment that also contains that fragment. If found, it uses additional processing (such as fine-grained alignment information) to identify a fragment of the target TU segment to propose as a translation of the query fragment. Each type of matching can be subject to a minimum length of matching fragment. Because some tokens are essentially valueless for recall (e.g. whitespace, tag tokens), the length is calculated by counting the number of 'feature tokens' in the fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.GetFeatures(System.Collections.Generic.List{Sdl.LanguagePlatform.Core.Tokenization.Token},System.Globalization.CultureInfo,System.Collections.Generic.List{System.Int16}). In addition, some match fragments may be of no interest for recall if they consist only of tokens with little translational significance, such as stop words. So, each type of matching can be further subject to a minimum number of significant feature tokens in a matching fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.IsSignificantForRecall(Sdl.LanguagePlatform.Core.Tokenization.Token)).

MinTM_TDBFeatures

The minimum length of a TM-TDB match fragment (see SubsegmentSearchSettings remarks)

Declaration
public int MinTM_TDBFeatures
Field Value
Type Description
Int32
Remarks

TM-TDB matching (aka 'baseline' subsegment recall) process fragments of the query segment, and for each fragment, searches for a complete source TU segment matching it. If found, the complete target TU segment is proposed as a translation of the query fragment. DTA matching (aka 'full' subsegment recall) process fragments of the query segment, and for each fragment, searches for a source TU segment that also contains that fragment. If found, it uses additional processing (such as fine-grained alignment information) to identify a fragment of the target TU segment to propose as a translation of the query fragment. Each type of matching can be subject to a minimum length of matching fragment. Because some tokens are essentially valueless for recall (e.g. whitespace, tag tokens), the length is calculated by counting the number of 'feature tokens' in the fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.GetFeatures(System.Collections.Generic.List{Sdl.LanguagePlatform.Core.Tokenization.Token},System.Globalization.CultureInfo,System.Collections.Generic.List{System.Int16}). In addition, some match fragments may be of no interest for recall if they consist only of tokens with little translational significance, such as stop words. So, each type of matching can be further subject to a minimum number of significant feature tokens in a matching fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.IsSignificantForRecall(Sdl.LanguagePlatform.Core.Tokenization.Token)).

MinTM_TDBSignificantFeatures

The minimum number of significant tokens in a TM-TDB match fragment (see SubsegmentSearchSettings remarks)

Declaration
public int MinTM_TDBSignificantFeatures
Field Value
Type Description
Int32
Remarks

TM-TDB matching (aka 'baseline' subsegment recall) process fragments of the query segment, and for each fragment, searches for a complete source TU segment matching it. If found, the complete target TU segment is proposed as a translation of the query fragment. DTA matching (aka 'full' subsegment recall) process fragments of the query segment, and for each fragment, searches for a source TU segment that also contains that fragment. If found, it uses additional processing (such as fine-grained alignment information) to identify a fragment of the target TU segment to propose as a translation of the query fragment. Each type of matching can be subject to a minimum length of matching fragment. Because some tokens are essentially valueless for recall (e.g. whitespace, tag tokens), the length is calculated by counting the number of 'feature tokens' in the fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.GetFeatures(System.Collections.Generic.List{Sdl.LanguagePlatform.Core.Tokenization.Token},System.Globalization.CultureInfo,System.Collections.Generic.List{System.Int16}). In addition, some match fragments may be of no interest for recall if they consist only of tokens with little translational significance, such as stop words. So, each type of matching can be further subject to a minimum number of significant feature tokens in a matching fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.IsSignificantForRecall(Sdl.LanguagePlatform.Core.Tokenization.Token)).

SubsegmentMatchTypes

The types of subsegment matching that should be used. See remarks for SubsegmentSearchSettings

Declaration
public HashSet<SubsegmentMatchType> SubsegmentMatchTypes
Field Value
Type Description
HashSet<SubsegmentMatchType>
Remarks

TM-TDB matching (aka 'baseline' subsegment recall) process fragments of the query segment, and for each fragment, searches for a complete source TU segment matching it. If found, the complete target TU segment is proposed as a translation of the query fragment. DTA matching (aka 'full' subsegment recall) process fragments of the query segment, and for each fragment, searches for a source TU segment that also contains that fragment. If found, it uses additional processing (such as fine-grained alignment information) to identify a fragment of the target TU segment to propose as a translation of the query fragment. Each type of matching can be subject to a minimum length of matching fragment. Because some tokens are essentially valueless for recall (e.g. whitespace, tag tokens), the length is calculated by counting the number of 'feature tokens' in the fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.GetFeatures(System.Collections.Generic.List{Sdl.LanguagePlatform.Core.Tokenization.Token},System.Globalization.CultureInfo,System.Collections.Generic.List{System.Int16}). In addition, some match fragments may be of no interest for recall if they consist only of tokens with little translational significance, such as stop words. So, each type of matching can be further subject to a minimum number of significant feature tokens in a matching fragment (see Sdl.Core.FineGrainedAlignment.Core.SubsegmentUtilities.IsSignificantForRecall(Sdl.LanguagePlatform.Core.Tokenization.Token)).