Difference between revisions of "LangID"
Jump to navigation
Jump to search
| Line 8: | Line 8: | ||
|- | |- | ||
| Percent Complete | | Percent Complete | ||
| − | | style="text-align:center;" | | + | | style="text-align:center;" | 48% |
|- | |- | ||
| Message Queue | | Message Queue | ||
Revision as of 11:16, 22 June 2021
| Executable | LangID.exe
|
| Stage | 18 |
| Percent Complete | 48% |
| Message Queue | langid
|
The LangID ETL scans ascii and unicode files and identifies non-english language. If it is over a certain confidence threshold it will then tag the document with the language it found.
File Types
LangID processes the following types of files.
| File Type | Produces |
|---|---|
| Type_ASCII_Text | Tags |
| Type_UTF8_Encoded_Text | Tags |
| Type_Extracted_ASCII | Tags |
| Type_Extracted_Unicode | Tags |
Note: The types Type_Extracted_ASCII and Type_Extracted_Unicode are typically provided by the TextExtract ETL.
Configuration
| Name | Data Type | Description |
|---|---|---|
| langid_min_probability | float | The minimum threshold for a positive language identification. Between 0.0 and 1.0. Default is 0.333. |
| langid_max_bytes_to_interpret | integer | The maximum bytes to scan for language identification. Default is 1000. |
E.G. for TruxtonSettings.xml
<truxton_options>
<!-- ...
other configs
...
-->
<langid_min_probability>0.75</langid_min_probability>
<langid_max_bytes_to_interpret>12000</langid_max_bytes_to_interpret>
</truxton_options>
