topic Re: When are noise characters going to be implemented? in Data Management Questions

When are noise characters going to be implemented?

JulieKanzler — Tue, 27 Sep 2011 16:45:23 GMT

I need to ignore spaces when matching alphanumeric house numbers (e.g., "101A" matches to "101 A"). It seems like the noise characters found in the XML configuration would be the ideal way of handling this. Am I correct? But I see based on the "Customizing..." white paper that they aren't yet implemented. Anyone know when they will be implemented? Or have ideas for a work-around? Thanks!

Re: When are noise characters going to be implemented?

JulieKanzler — Tue, 27 Sep 2011 17:51:26 GMT

hmmm... in doing some testing, the noise characters do appear to be implemented (for some fields, at least).

Re: When are noise characters going to be implemented?

BradNiemand — Wed, 28 Sep 2011 15:30:33 GMT

What style are you trying to build a locator with? Single House, Dual Ranges?

Brad

Re: When are noise characters going to be implemented?

ThibautDelaere1 — Fri, 30 Sep 2011 06:23:12 GMT

hmmm... in doing some testing, the noise characters do appear to be implemented (for some fields, at least).

In the standard US address style they seem only implemented for the 'Street' field. You could alter the address style and make them work for other fields.

Re: When are noise characters going to be implemented?

JulieKanzler — Fri, 30 Sep 2011 11:32:28 GMT

ThibautD - you are correct; they are only ignored in the street column. If you have a space in the address number, the user must also enter a space to match it. If the reference data leaves out the space, an address with a space in the house number won't match. Seems inconsistent with the concept of geocoding. And the old 9.3 geocoders had no problem with this. Brad- I am trying to customize the US Single Address locator. Any ideas?

Re: When are noise characters going to be implemented?

JulieKanzler — Fri, 30 Sep 2011 11:47:40 GMT

Oh and ThibautD, you suggested fixing the problem by editing the style, and I am trying to do just that. But it's not clear how when the noise_chars don't appear to be field-specific. They are just under grammar>spelling>character_equivalency_table. Thanks for your help.

Re: When are noise characters going to be implemented?

BradNiemand — Fri, 30 Sep 2011 15:31:31 GMT

Do you have any of the Service Packs installed?

Brad

Re: When are noise characters going to be implemented?

JulieKanzler — Fri, 30 Sep 2011 16:44:35 GMT

I'm at ArcGIS 10 service pack 2

Re: When are noise characters going to be implemented?

BradNiemand — Fri, 30 Sep 2011 19:26:45 GMT

First things first, you are correct, spelling tables only apply to street name. We will work to apply these to more fields in SP3.

Unfortunately, if you would like to do this to house number there are a few changes that will need to be made.

First go here:

<mapping_schema name="SingleAddress" geom_type="point">

Under this mapping_schema look for

<field name="StreetName" grammar_ref="StName">

You will find the following section:

            <scoring_method ref="calculate_score">
              <init_properties>
                <prop name="CharacterTable">scoring</prop>
              </init_properties>
              <parameter>
                <input_value />
              </parameter>
              <parameter>
                <field_value ref="StreetName" />
              </parameter>
            </scoring_method>

This is how you apply spelling to other components of the mapping_schema. Something like this:

          <field name="House" grammar_ref="House">
            <desc>House number</desc>
            <preferred_name>HN</preferred_name>
            <preferred_name>ADDRESS</preferred_name>
            <scoring_method ref="calculate_score">
              <init_properties>
                <prop name="CharacterTable">scoring</prop>
              </init_properties>
              <parameter>
                <input_value />
              </parameter>
              <parameter>
                <field_value ref="House" />
              </parameter>
            </scoring_method>
          </field>

As for House Number though, it is indexed for the Single House Style so in addition to the changes above, you will need to change the reference in the reference_data_style section as well.

This is the start of that section. Search for "Single House (Address Points)":

    <ref_data_styles>

      

      <ref_data_style>
        <name>Single House</name>
        <desc>US Single House Addresses</desc>
        <table_roles>
          <table_role name="Primary">
            <display_name>Primary Table</display_name>
            <desc>Address feature class</desc>
            <field_roles>
              <field_role name="Primary.Shape" is_geometry="true">
                <display_name>Geometry</display_name>
                <preferred_name>Shape</preferred_name>

Now you will need to change the mapping schema reference from SingleAddressSearch to SingleAddressPolygonCentroid.

        <data_source type="indexed">
          <mapping_schema ref="SingleAddressSearch" />

to

        <data_source type="indexed">
          <mapping_schema ref="SingleAddressPolygonCentroid" />

This should do it but I may have forgot something. Let me know how it works.

Brad

Re: When are noise characters going to be implemented?

JulieKanzler — Mon, 03 Oct 2011 11:48:40 GMT

Thanks, Brad - that helped! Just one question - is there a data dictionary or other reference for the XML format used in the locator configuration file. I found the "Customizing Locators in ArcGIS 10" whitepaper and assumed that was it, but there's nothing in there documenting these different mapping schemas. The comments in the xml file help, but I still don't have a sense of when I would point to one versus another.

Thanks again!

Re: When are noise characters going to be implemented?

JulieKanzler — Mon, 03 Oct 2011 14:46:20 GMT

Thanks, Brian! That was really helpful. Is there a reference where I can read about all of the XML objects and attributes used in the locator configuration file? Or at least some kind of data dictionary? I can kind of infer a bit from the names and the comments, but I have no idea (for example) when you might use one mapping_schema vs. another one. And why is this one called "...PolygonCentroid"? That is not the case for our reference data, and why should it matter for matching purposes anyway? I have tons of questions when I look at the configuration files, and just wonder if there's something I can do to understand better, without needing to reverse engineer. Thanks so much, once again.

FYI - for others reading with the same issue - to completely fix this, I needed to add some alternatives to <def name="AlphaNumericHouse">, most importantly: <elt ref="number" /><elt ref="alpha"/>

Re: When are noise characters going to be implemented?

BradNiemand — Mon, 03 Oct 2011 16:35:14 GMT

The only doc we have for customizing and understanding the locator xml is the "Customizing Locators in ArcGIS 10". We will work on enhancing this in the future.

Brad

Re: When are noise characters going to be implemented?

JulieKanzler — Mon, 03 Oct 2011 16:39:07 GMT

Yep, I read that of course. Anyone who customizes this file at least needs a data dictionary. Thanks again.

Re: When are noise characters going to be implemented?

DavidWarnock — Wed, 03 Sep 2014 08:42:53 GMT

I wanted to resurrect this post because it's been 2 years and 11 months without any sign of better documentation. "Customizing Locators in ArcGIS 10" still claims that Noise Characters are not yet implemented. Better documentation for this is a must-have. As far as I can see, there has been no new documentation or revisions of existing documentation for nearly 4 years (the white paper is dated Nov 2010). That would be fine if the system hadn't changed, but it's obvious it has. Those of us using custom datasets or non-US address formats are basically rooting around in the dark, taking guesses at what various things do.

-David

Re: When are noise characters going to be implemented?

ThibautDelaere1 — Wed, 03 Sep 2014 08:59:40 GMT

Having to edit the locator styles clients needs on regular basis it would be great to have a better (or at least) updated reference document.

Also, edits that work fine in one version of ArcGIS, do not always work in other versions. There is no documentation at all about this.