Datacleanse name parsing for the English dictionary

Question

I am using the person_firm_en (English) dictionary for name parsing (Datacleanse for Data Services XI) for Canada data. There are many situations where the full name consists of 3 names and datacleanse parses the 2nd and the 3rd name into the 'family name' field. Thus, instead of parsing the 2nd name into the given_name2 field, it puts it with the last name into the family_name field. Some examples are:
Example 1 - ROBERT GRAHAM LEITCH
given_name1 - Robert
family_name - Graham Leitch
Example 2 - JEREMY HURON NEAL
given_name1 - Jeremy
family_name - Huron Neal
SAP's response to this issue is that each entry which is not being parsed out as given_name2 needs to have the person_firm_en dictionary entry edited to add it as a given name. This can also impact its parsing if this name appears as a last name.
Does anyone have any other suggestions or any rule/code that you have implemented to handle such situations?

Answer 1

Shenaz,
We have struggled quite a bit with the name parsing n the Data Cleanse (especially when using version 12.1.1). I wrote some custom functions that do name parsing and in many cases, especially for middle names, they seem to do a better job. Below is the code that returns Person1 Middle name from a passed in mixed name value. The functions CF_Name_Standardize_Replace_Chars and CF_Remove_Extra_Spaces contain find\replace tables and remove things like punctuation, MR, MRS, empty space, etc.
1 Have 5 other functions, to pull out First, Middle, and Last names of persons 1 and 2. They look very similar, the only different part being the last section where the word pattern is analyzed and the decision is made as to which word is to be returned.
I could get you code for the others if it would help.
#prep
$return_string = '';
$word_1 = '';
$word_2 = '';
$word_3 = '';
$word_4 = '';
$word_5 = '';
$word_6 = '';
$word_7 = '';
$word_count = 0;
$working_string = ltrim_blanks ( rtrim_blanks( $input_string ) );
$working_string = upper( $working_string );
$working_string = CF_Name_Standardize_Replace_Chars( $working_string );
if ( nvl ( index ( $working_string, ',', 1), 0 ) = 0)
begin
  $has_comma = 'false';
end
else
begin
  $has_comma = 'true';     
end
#assign the words to their variables
if( $has_comma = 'true' )
begin
  $split_index = index ( $working_string, ',', 1);
  $word_1 = ltrim_blanks(rtrim_blanks(substr( $working_string, 1, $split_index - 1 )));
  $word_1 = CF_Remove_Extra_Spaces( $word_1 );
  $word_1 = replace_substr( $word_1 , ' ', '-' ); #we will assume it is a last name and should be hyphenated
  $working_string = substr( $working_string, $split_index + 1, 255 );
  $working_string = ltrim_blanks( rtrim_blanks( replace_substr( $working_string , ',', ' ' ) ) );
  $working_string = CF_Remove_Extra_Spaces( $working_string );
  #these are now offset by one because we removed the first "word" before the comma.
  $word_2 = nvl( word( $working_string, 1 ), '');
  $word_3 = nvl( word( $working_string, 2 ), '');
  $word_4 = nvl( word( $working_string, 3 ), '');
  $word_5 = nvl( word( $working_string, 4 ), '');
  $word_6 = nvl( word( $working_string, 5 ), '');
  $word_7 = nvl( word( $working_string, 6 ), '');
end
else
begin
  $working_string = CF_Remove_Extra_Spaces( $working_string );
  $word_1 = nvl( word( $working_string, 1 ), '');
  $word_2 = nvl( word( $working_string, 2 ), '');
  $word_3 = nvl( word( $working_string, 3 ), '');
  $word_4 = nvl( word( $working_string, 4 ), '');
  $word_5 = nvl( word( $working_string, 5 ), '');
  $word_6 = nvl( word( $working_string, 6 ), '');
  $word_7 = nvl( word( $working_string, 7 ), '');
end
#figure out what the word count is
if ($word_1 != '')
begin
  $word_count = 1;
end
if ($word_2 != '')
begin
  $word_count = 2;
end
if ($word_3 != '')
begin
  $word_count = 3;
end
if ($word_4 != '')
begin
  $word_count = 4;
end
if ($word_5 != '')
begin
  $word_count = 5;
end
if ($word_6 != '')
begin
  $word_count = 6;
end
if ($word_7 != '')
begin
  $word_count = 7;
end
#and finally, analyze the word pattern and make a decision where the desired word is
if ($has_comma = 'true')
begin
  if ($word_count = 6 AND $word_4 = '&')
  begin
    $return_string = $word_3;
  end
  if ($word_count = 5 AND $word_4 = '&')
  begin
    $return_string = $word_3;
  end
  if ($word_count = 4 AND $word_3 != '&')
  begin
    $return_string = $word_3 || '-' || $word_4; #assume a middle name that should be hyphenated in this case?? i.e. Smith, Damian John-Doe (From Smith, Damian John Doe) - (Return John-Doe)
  end
  if ($word_count = 3)
  begin
    $return_string = $word_3;
  end
end
else
begin
  if ($word_count = 3 AND $word_2 != '&')
  begin
    $return_string = $word_2;
  end
  if ($word_count = 4 AND $word_2 != '&' AND $word_3 != '&' AND $word_4 != '&')
  begin
    $return_string = $word_2;
  end
  if ($word_count > 4 AND $word_3 = '&')
  begin
    #this stuff just in case we have a middle "initial" in the second position. In that case we will assume a middle name is there
    if (length( $word_2 ) = 1)
     begin
      $return_string = $word_2;
    end
    if (length( $word_2 ) = 2 AND index($word_2, '.', 1) = 2)
     begin
      $return_string = $word_2;
    end
    #end of middle "initial" search.
  end
  if ($word_count > 5 AND $word_4 = '&')
  begin
    $return_string = $word_2;
  end
end
Return init_cap( $return_string );