Articles in this section

See more

A Quick Look at RNI-RNT: Running a Sample Program

Avatar
Jason Stevens
  • Updated
Follow

Building and Running the Sample Applications

To build and run the sample applications, you must have the Java SDK (11 or later). To use the Ant build files we provide to build and run the samples, you need Ant (1.7.1 or later) with the JAVA_HOME environment variable set to the root of your Java SDK. For more information, see http://ant.apache.org.

The source files for these applications and the Ant build file for compiling and running them (build.xml) are located in $BT_ROOT/rlpnc/samples/java.

Tip

The Ant scripts and build files require one input property: bt.arch=$BT_BUILD (bt.arch=amd64-glibc217-gcc48, for example). If you set this property in the script (build.xml), you do not need to include it on the command line.

Table 2. Sample Applications

Source File

Description

AddNamesSample.java

Adds names from a UTF-8 file to an RNI Index.

LoadGazetteerSample.java

Loads an XML gazetteer into an RNI Index.

IndexQuerySample.java

Submits a series of queries (names) to an index and reports on the results.

DistributedTransactionSample.java

Queries an index, deletes the names returned from that index, and adds the names to a second index. The deletions and additions are performed in a single distributed transaction with two-phase commit.

MatchNamesSample.java

Determines the similarity of two or more names.

MatchPhenomenaSample.java

Demonstrates the different name matching phenomena that RNI supports.

AutomatedTranslationSample.java

Translates one or more names.

InteractiveTranslationSample.java

Simulates a series of user interactions resulting in the translation of an Arabic name.

RNISolrjSample.java

Integrates RNI with Solr to add and query Solr documents with multiple and multivalued name fields.

AddressIndexQuerySample.java

Submits a series of queries (addresses) to an index and reports on theresults.

AddressMatchPhenomenaSample.java

Demonstrates the different address matching phenomena that RNIsupports.

Your License

You must copy the license file you obtained from BasisTech to $BT_ROOT/rlp/rlp/licenses. If the license is not in place, you cannot access any RNI-RNT functionality. The license defines the scope of the activities you may perform with RNI-RNT.

Using the Ant Build Script

Tip

The Ant scripts and build files require one input property: bt.arch=$BT_BUILD (bt.arch=amd64-glibc217-gcc48, for example). If you set this property in the script (build.xml), you do not need to include it on the command line.

Change directory to $BT_ROOT/rlpnc/samples/java and run Ant:

ant -Dbt.arch=$BT_BUILD target

where target is one of the Ant build targets in the following table.

target

Description

compile

Compiles the samples and places the class files in $BT_ROOT/rlpnc/samples/java/obj/$BT_BUILD.

compile.Class[a]

Compiles the specified sample.

run

Compiles (if necessary) and runs the samples with the command-line arguments defined in the Ant build file. Each sample prints a message to the console indicating what it has done, including any file it has created.

run.Class [a]

Runs the specified sample.

clean

Removes the class files and any files created by the samples.

clean.Class [a]

Removes the sample class file(s) and any file created by the sample.

all

Calls compile and run.

[a] Class is the sample class name. Use the Class targets to compile, run, or clean a single sample. For example, to run LoadGazetteerSample, the target is run.LoadGazetteerSample.

As you create your own applications, you can use the Ant build file as the starting point for establishing your own build procedures.

RNI Command-Line Interface

RNI includes a Java command line interface RNICLI. This is provided in binary form only in btrlpnc.jar.

Use RNICLI to create RNI indexes, to add names from gazetteers or UTF-8 flat files, to run RNI queries, and to pair-wise match in a query file against names in an input file.

Use the command-line utility script in rlpnc/samples/java/bin. For Unix, the script is rnicli.sh. For Windows, the script is rnicli.bat. Include the arguments shown below.

Tip

Windows Powershell users

The Windows Powershell command line does not automatically use UTF-8 encoding. To use the command-line interface with non-Latin script languages:

  1. Set your system locale to UTF-8

  2. Add this flag to each command:

    -Dfile.encoding=65001

Table 3. RNICLI Commands

Description

Command

Create a new index

RNICLI create -root -index [-in] [-lang] [-entity] [-threads] [ignoreBadData] [-langOfOrigin]

Add names to an existing index

RNICLI add -root -index -in [-lang] [-entity] [-threads] [ignoreBadData] [-langOfOrigin]

Query an index

RNICLI query [ignoreBadData] -root -index -query [-report] [-lang] [-entity] [-threads] [ignoreBadData] [-threshold] [-maxToConsider][-scoreToCheckRestriction] [-langOfOrigin] [-maxToCheck] [-namesToCheckAllowance] [-max]

Pair-wise match

RNICLI match -root -in -query [-report] [-lang] [-entity] [-threads] [ignoreBadData] [-langOfOrigin] [-threshold]

For more information on the command-line interface, including definitions of the arguments, display the help using one of the following commands:

RNICLI -?
RNICLI -help

Using the Ant Build File

You can also use the Ant build file in $BT_ROOT/rlpnc/samples/java to run the RNI command-line interface.

Tip

The Ant scripts and build files require one input property: bt.arch=$BT_BUILD (bt.arch=amd64-glibc217-gcc48, for example). If you set this property in the script (build.xml), you do not need to include it on the command line.

Change directory to $BT_ROOT/rlpnc/samples/java and run Ant:

ant -Dbt.arch=$BT_BUILD args target

where args provide argument values and target is one of the Ant build targets in the following table. The argument values are described in the RNICLI help file. For any arguments you omit, the Ant build file supplies default values. You can also edit build.xml to change these values.

Optional args

target

Description

 
-Drni-index=index_path
-Drni-source=in_file
-Drni-lang=lang
-Drni-entity=entity_type
-Darg=ignoreBadData

RNICLI.create

Creates an index from a gazetteer or UTF-8 flat file.

 
-Drni-index=index_path
-Drni-add=in_file
-Drni-lang=lang
-Drni-entity=entity_type
-Darg=ignoreBadData

RNICLI.add

Adds names to an existing index from a gazetteer or UTF-8 flat file.

 
-Drni-index=index_path
-Drni-query=query_file
-Dmatch-lang=lang 
-Drnicli-report=report_file
-Drni-entity=entity_type
-Drnicli-threshold=min_score
-Drni-max-return=max_return                    
-Darg=ignoreBadData

RNICLI.query

Runs a series of queries against an index and reports the results.

RNICLI.query.OFAC

Calls RNICLI.query with the OFAC_SDN[a] index and OFAC-queries.txt

RNICLI.query. JapaneseBaseballPlayers

Calls RNICLI.query with the indexJapaneseBaseballPlayers index and baseball_queries_jpn.txt query file. The index contains the names of Japanese professional baseball players (some from other countries) in various Japanese scripts. You can modify the query file, which contains players names in Latin script.

 
-Dindex-names-in=in_file
-Drni-query=query_file
-Dmatch-lang=lang 
-Drnicli-report=report_file
-Drni-entity=entity_type
-Drni-threshold=min_score

RNICLI.match

Pair-wise match names in an input file against names in a query file and reports the results.

 
-Drni-index=index_path
-Drni-report=report_file

RNICLI.clean

Removes the RNICLI report file.[b]

 
-Drni-index=index_path
-Drni-source=in_file
-Drni-query=query_file
-Dlang=lang
-Drnicli-report=report_file

RNICLI

Calls RNICLI.clean, RNICLI.create, RNICLI.query, and RNICLI.generate-keys

[a] OFAC_SDN contains the names in Latin script from the Office of Foreign Asset Control watch list. OFAC-queries.txt contains query names. To run your own queries, you can modify this file.

[b] If an RNI index already exists when you run RNICLI.create, it is overwritten. If you want to remove it, do so manually.

RNT Command-Line Interface

The RNT command-line interface, RNTCLI , translates names from one text domain to another, as described in Translating Names. For each name in an input file, RNTCLI generates a translation and a confidence score (0 - 1) associated with the translation.

Use the command-line utility script in rlpnc/samples/java/bin. For Unix, the script is rntcli.sh. For Windows, the script is rntcli.bat. Include the arguments shown below.

Tip

Windows Powershell users

The Windows Powershell command line does not automatically use UTF-8 encoding. To use the command-line interface with non-Latin script languages:

  1. Set your system locale to UTF-8

  2. Add this flag to each command:

    -Dfile.encoding=65001

Table 4. RNTCLI Arguments

Argument

Description

root

$BT_ROOT directory; always required

in

path to the input file; always required

targTr

transliteration for the target language; only required if not specified in the input file

entity

entity type of the names in the infile, set to PERSON if omitted

srcScr

script for the source, set to the default script if omitted.

srcLang

language of the names in the input file, statistically guessed if omitted

srcTr

transliteration scheme for the source, set to native if omitted

targScr

script for the target, set to default script if omitted

targLang

target language, set to eng if omitted

threads

number of threads to use for translation, 1 if omitted

results

maximum number of results returned, 16384 if omitted

tokens

maximum number of tokens to allow in Name data, 10 if omitted

report

path to the report file, report to console if omitted

langOfOrigin

language of origin, statistically guessed if omitted

O=

flags, where valid options are:

-Oortho=yes|no

-Ostat=yes|no

-Operftrade=normal|fast|careful|precise

-Osegment=yes|no

-OvariantSpelling=yes|no

-Oregion=default|north|south

-OkorGeography=default|northKorean|southKorean

-Onormalize=yes|no

For more information on the command-line interface, display the help using one of the following commands:

RNTCLI -?
RNTCLI -help

Using the Ant Script

You can also use the Ant script in $BT_ROOT/rlpnc/samples/java to run the RNT command-line interface.

Tip

The Ant scripts and build files require one input property: bt.arch=$BT_BUILD (bt.arch=amd64-glibc217-gcc48, for example). If you set this property in the script (build.xml), you do not need to include it on the command line.

Change directory to $BT_ROOT/rlpnc/samples/java and run Ant:

ant -Dbt.arch=$BT_BUILD args target

where args provide argument values and target is one of the Ant build targets in the following table.

Optional args[a]

target

Description

-Drntcli-in=in-file -Drntcli-report=report-file -DsrcLang=src-lang-code -DsrcScr=src-scr -DsrcTr=src-translit -DtargLang=targ-lang-code -DtargScr=targ-scr -DtargTr=targ-translit -Drnt-entity=entity_type

RNTCLI

Translates the names in the input file

-Drntcli-report=report-file

RNTCLI.clean

Removes the RNTCLI report file.

[a] Where in-file, report-file, src-lang-code, src-scr, src-translit, targ-lang-code, targ-scr, targ-translit, and entity_type are as described in the previous section. For any arguments you omit, the Ant script supplies default values. You can also edit build.xml to change these values.

Related articles

Comments

0 comments

Article is closed for comments.





Powered by Zendesk