This update contains several important bug fixes in the core Salience library and supporting data directory. These bug fixes are in the areas of document categories, named entity relationships and opinions, and intentions, as well as bug fixes to the Windows demo app. More details are provided below.
IMPORTANT: Access to intention functionality is controlled via the license file. Although Salience 5.x license files can be used with Salience 6, existing customers will need to contact Lexalytics for updated license files in order to access intention functionality.
· Fix to data file issue in the core engine which caused failure in the API call to lxaGetDocumentCategories (and related wrapper functions) on Linux platforms.
· Fix in core engine to resolve issue in getting document details when using "fail on long sentences" option.
· Fix to data file issue in the core engine which resulted in decreased extraction of job titles, affecting relationship output.
· Fix to data file issue in the core engine which resulted in decreased pronominal linking, affecting entity extraction, relationship, and opinion output.
· Fix to data file to insert quotes into Date label, allowing "Date" to appear as entity type in named entity output.
· Fix in Windows Demo app to bug in persisting modification to license file path in Salience demo app options.
· Updated SalienceDemo source code solution in Windows distribution with additions to demo app for Salience 6.
· Updating Python, Java, .Net examples to include calls to Document Categories.
· Fix in Windows demo app to modify storage of demo options settings, avoids error in Salience 6 demo app attempting to use Salience 5.x demo app settings.
· Fix in Windows demo app to error in Salience 6 document classification options when initializing a session with a non-English data directory.
· Fix to company entity CDL files in default data directory that led to excess low precision company entity extractions.
· Improved Intention restriction files so that you can have separate rules for positive and negative instances. Previously the rule for "not buying here again"=quit was overlapping with the rule for "buying a car" so that "not buying a car today" became a quit.
· Modification to the core engine to allow facet rollup in a case-insensitive manner.
· Modification to .Net wrapper to pass through errors from core library, allowing display of error message when attempting to extract intentions with a license not authorized for intentions.
· Modification to POS tagger data file whitelist "cannot" as an adverb to fulfill use as negator.
· SentimentTuner: The Sentiment Tuning tool is a command-line tool for analyzing the performance of sentiment dictionaries for specific domains of content. More information about how to use this tool and its output will be posted to the Lexalytics Dev Wiki.
If you have any questions, please email [email protected]
Updated over 1 year ago