The Hekate Markup Language (HML) is provided for the machine readable XML serialization of the HeKatE rule base described in a human-readable HMR format.

The rules are formalized with the use of the ALSVFD logic and XTT2 rules prototyped with the ARDplus method.

There are the following subsets of the HeKatE Markup Language defined:

• attml – Attribute Markup Language for describing rule attributes
• ardml – Attribute Relationship Markup Language for the ARDplus prototype
• xttml – XTT2 Rule Markup Language for the XTT2 structured rule representation

The sublanguages can be used according to the following scenarios:

1. attml – only attribute definition, the minimalistic case
2. attml + ardml – attributes and ARD dependencies
3. attml + ardml + xttml – attributes, dependencies and XTT2 rules
4. attml + xttml – attributes and rules, (without the ARD prototype)

A HML application using all sublanguages is as follows:

<hml>
  <!-- attml starts here -->
  <types>
    ...
  </types>
 
  <attributes>
    ...
  </attributes>
  <!-- attml ends here -->
 
  <!-- ardml starts here -->
  <properties>
    ...
  </properties>
 
  <tph>
    ...
  </tph>
 
  <ard>
    ...
  </ard>
  <!-- ardml ends here -->
 
  <!-- xttml starts here -->
  <xtt>
    ...
  </xtt>
  <!-- xttml ends here -->
 
  <!-- extra system state specification -->
  <states>
  </states>
</hml>

The current version is referred to as HML2 (the first one is HML1, see below).

The markup for the various knowledge representations now present in HeKatE has been evolving.

• 2004 - first markup for XTT in GJN PhD → XTTML0
• 2007 - first markup for ARD+ in VARDA
• 2007 - second version of markup for XTT in early versions of HQEd → XTTML
• 2007 - third version of markup for XTT in HQEd M5_* → XTTML2
• 2008 - first version markup proposal for HeKatE, ARD+, XTT+ → HML1: ATTML1, ARDML1, XTTML3 (proposal for XTT+, partially supported by design tools)
• 2009 - second version of HML, ARD+ XTT2 → HML2: ATTML2, ARDML2, XTTML4 (supported by HQEd M6_* from 05.2009)

HeKatE markup language DTD: (for HML2).

For syntax test purposes there is an example XML file:hml-test-hekate_case_thermostat.xml A minimalistic file is: hml-test-min.xml

(Note: the above files are versioned in the HaDEs CVS repo in the HatHoR module.)

Important elements are indexed with an attribute id. id value has to be unique (enforced by DTD) and should start with (depending on particular element):

• types: id=“tpe_…”
• attributes: id=“att_…”
• properties: id=“prp_…”
• type groups: id=“tgr_…”
• attribute groups: id=“agr_…”
• dependencies: id=“dep_…”
• ARD history (TPH): id=“hst_…”

All id numbers start with 1

An additional attribute (id) is introduced to guarantee that types, attributes, properties and groups can be uniquely referenced. It can be forced, using DTD, that an attribute has a unique value within an XML document, it becomes a unique identifier. Similarly, it can be forced that an attribute references an element by its unique identifier. Making attribute names unique identifiers prevents naming a group, or type, with the same name as the attribute, since unique identifiers are unique within entire document regardless of the element they regard to.

It is possible to validate a HML file using a validator i.e. xmlstarlet:

xmlstarlet val -e -d hml.dtd hml-test.xml

For information on attributes see here.

  <types>
    <type id="tpe_1" name="Temperature" base="numeric" length="5" scale="0">
      <desc>represents a temperature</desc>
      <domain>
        <value from="1" to="5"/>
        <value is="7" />
        <value from="20" to="22"/>
        <value is="26" />
        <value is="27" />
      </domain>
    </type>
    <type id="tpe_2" name="Integer" base="numeric" length="5" scale="0">
      <desc>represents an integer number</desc>
      <domain>
        <value from="-10000" to="10000"/>
      </domain>
    </type>
  </types>
 
  <attributes>
    <attr id="att_1" type="tpe_1" name="temp" abbrev="t1" class="simple" comm="in">
      <desc>input temperature from sensor number 1</desc>
    </attr>
    <attr id="att_2" type="tpe_1" name="temp-aux" abbrev="t1a" class="general" comm="in">
      <desc>input temperature from sensor number 1, aux reading, including historic data</desc>
    </attr>
 
    <agroup id="grp_1" name="sensors" abbrev="sns">
      <desc>temperatures from sensors</desc>
      <attref ref="att_1"/>
      <attref ref="att_2"/>
    </agroup>
 
  </attributes>

Both in the case of types and attributes groups can be specified. Example: types day, hour, month can form a type group date. The same could apply to specific attributes. In the example above there is set of attributes describing sensor readings formed out of values of temp and corresponding temp-aux. Attributes in groups and types are ordered.

For ARD see ardplus and gjn2008flairs-ardformal gjn2008flairs-ardprolog

<hml>
  <types>
    ...
  </types>
 
  <attributes>
    <attr name="Thermostat" id="att_0"/>
    <attr name="Time" ... />
  </attributes>
 
  <properties>
    <property id="prp_1">
      <attref ref="att_0"/>
    </property>
    <property id="prp_2">
      <attref ref="att_0"/>
      <attref ref="att_1"/>
    </property>
    <property id="prp_3">
      <attref ref="att_0"/>
    </property>
    <property id="prp_4">
      <attref ref="att_1"/>
    </property>
    <property id="prp_5">
      <attref ref="att_0"/>
    </property>
  </properties>
 
  <tph>
    <hist id="hst_01" src="prp_1" dst="prp_2"/>
    <hist id="hst_02" src="prp_2" dst="prp_3"/>
    <hist id="hst_03" src="prp_2" dst="prp_4"/>
    <hist id="hst_04" src="prp_4" dst="prp_5"/>
  </tph>
 
  <ard>
    <dep id="dep_01" independent="prp_1" dependent="prp_2"/>
    <dep id="dep_02" independent="prp_2" dependent="prp_3"/>
    <dep id="dep_03" independent="prp_3" dependent="prp_4"/>
    <dep id="dep_04" independent="prp_2" dependent="prp_4"/>
  </ard>
 
  ...
  ...
</hml>

Basic syntax:

<properties>
          <property id="prp_1">
               <attref ref="att_0"/>
               ...
               <attref ref="att_ad003"/>
          </property>
		...
		<property id="prp_2">
               <attref ref="att_44"/>
               ...
               <attref ref="att_2"/>
          </property>
</properties>

• Properties are defined in <properties>…</properties> section. This section must contain all properties which are being used in an ARD diagram.
• Single property is defined by <property> tag
• Single property contains any number of attributes. A list of attributes is defined by <attref>

Basic syntax:

     <tph>
          <hist src="prp_1" dst="prp_2"/>
          <hist src="prp_2" dst="prp_3"/>
          <hist src="prp_2" dst="prp_4"/>
		...
          <hist src="prp_n" dst="prp_m"/>
     </tph>

• TPH diagram is definied in <tph>…</tph> section. TPH diagram contains relationships called transformations.
• A transformation is defined with the <hist> tags and it describes the development between a source (src) and destination (dst) properties.

Basic syntax:

<ard>
     <dep independent="prp_1" dependent="prp_2"/>
     <dep independent="prp_2" dependent="prp_3"/>
     ...
     <dep independent="prp_x" dependent="prp_y"/>
     ...
     <dep independent="prp_x" dependent="prp_z"/>
</ard>

• ARD dependencies begin with the <ard> tag. The tag does not have attributes. Each dependence is defined with a a pair of properties: an independent and a dependent one.

For information on the XTT itself see here.

  <xtt>
    <table id="tab_1" name="first">
      <schm>
       <precondition>
         <attref ref="att_1"/>
       </precondition>
       <conclusion>
         <attref ref="att_0"/>
       </conclusion>
      </schm>
      <rule id="rul_1">
       <condition>
         <relation name="in">
           <attref ref="att_0"/>
           <set>
             <value from="1" to="5"/>
             <value is="8"/>
           </set>
         </relation>
       </condition>
       <decision>
          <trans>
           <attref ref="att_1"/>
            <expr name="add">
             <attref ref="att_1"/>
            <value is="1"/>
           </expr>
          </trans>
          <trans>
           <attref ref="att_1"/>
            <expr name="add">
              <value is="5"/>
              <expr name="sin">
                <attref ref="att_0"/>
             </expr>
           </expr>
          </trans>
       </decision>
       <link>
         <tabref ref="tab_1"/>
       </link>
      </rule>
    </table>
  </xtt>

An XTT diagram (xtt element) consists of some number of XTT tables (the table element). The above example can be read as:

xtt_1:
rule_1: if att_0 in <1,5>u{8} then att_1 = att_0 + 1 and att_1 = 5 + sin(att_0)

Element schem describes precondition and conclusion attributes that are allowed to be used in a particular XTT table. The schema is generated by the ARD→XTT transition algorithm directly from ARD.

A rule (rule) consists of a condition (condition) part and a decision part (decision). The condition is composed of the ALSVFD formulas, that specify certainrelations between attribute values and given set of values.

The decision part is decomposed into two parts:

• state transistion that can specify new state (attribute values), and
• external action to be executed; it is assumed it does not infulences the state in any way.

Each rule can explicitly transef the inference to a next table (possibly also a rule).

Number of named system states can also be specified. A single state is a vector of attribute values.

<states>
  <state id="sta_1" name="start">
    <attref ref="att_0"/>
    <set>
      <value from="1" to="5"/>
      <value is="8"/>
    </set>
    <attref ref="att_1"/>
    <set>
      <value from="1" to="5"/>
    </set>
  </state>
</states>

HML is used to serialize the knowledge in the HeKatE project in a machne redable format. It is used to interchange knowledge between the tools forming the HaDEs framework.

HML syntax has been designed to:

• correspond to the logical system formulation with XTT2 based on the ALSVfd logics, and
• match the syntax of HMR (the HeKatE Meta Representation) as supported by HeART - the HeKatE runTime environment.