There are several problems about Chinese collective nouns: 1. cannot distinguish some collective nouns with the same pronunciations; 2. cannot detect errors about some collective nouns for describing plural objects; 3. cannot detect the inappropriate collective nouns for depicting natural landscapes.
I made several corrections as below in XML forms and hopefully they are helpful in improving the quality of LT in Chinese language error detection.
<!-- Chinese rule, 2020-06-24 -->
<rule id="" name="颗与棵">
<pattern>
<marker>
<token>棵</token>
<token>糖</token>
</marker>
</pattern>
<message>量词<suggestion>棵</suggestion>不能修饰<suggestion>糖</suggestion></message>
<example correction=''>一<marker>棵糖</marker></example>
<example>一颗糖</example>
</rule>
<!-- Chinese rule, 2020-06-24 -->
<rule id="" name="只与支">
<pattern>
<token>只</token>
<token regexp='yes'>笔|铅笔|钢笔|水笔|毛笔|蜡笔|水彩笔</token>
</pattern>
<message>量词“只”不能修饰“笔”</message>
<example correction=''>一<marker>只笔</marker></example>
<example>一支笔</example>
</rule>
<!-- Chinese rule, 2020-06-24 -->
<rule id="" name="批与群">
<pattern>
<token>批</token>
<token regexp='yes'>羊|山羊|绵羊|公羊|牧羊|母羊|牡羊</token>
</pattern>
<message>量词“批”不能修饰“羊”</message>
<example correction=''>一<marker>批羊</marker></example>
<example>一群羊</example>
</rule>
<!-- Chinese rule, 2020-06-24 -->
<rule id="" name="堆与群">
<pattern>
<token>堆</token>
<token regexp='yes'>人|男人|女人|中国人|外国人|美国人|英国人|法国人</token>
</pattern>
<message>量词“堆”不能形容“人”</message>
<example correction=''>一<marker>堆人</marker></example>
<example>一群人</example>
</rule>