Subject elision

In this section, we show you another way to use lex rules for planning multiple sentences. This time, we show you how to combine sentences that have the same subject. For example:

When these sentences are joined with subject elision, the result is:

The two sentences have been combined with "and". The subject of the second sentence has been left out (or “elided”). We say the sentences have been joined with subject elision.

We’ll go through an example. The JSON data we’ll use is the same as for the last section:

{
    "employees": [
        {
            "name": "Peter Jones",
            "job": "Head of Innovation",
            "teamSize": 1,
            "location": "New York"
        },
        {
            "name": "George Scott",
            "job": "Head Janitor",
            "teamSize": 2,
            "location": "New York"
        },
        {
            "name": "Anna Smith",
            "job": "Head of Marketing",
            "teamSize": 12,
            "location": "New York"
        },
        {
            "name": "Sue Collins",
            "job": "Chief Software Developer",
            "teamSize": 15,
            "location": "London"
        }
    ]
}

But this time, we’ll use the two lex rules shown below:

<S id="jobId" classes="jobClass">
    <Subj word="{{msg.name}}">
        <Feature name="elidable" value="true"/>
    </Subj> 
    <VP word="be"/>
    <Obj word="{{msg.job}}"/> 
</S>
  
<S id="teamId" classes="teamClass">
    <Subj word="{{msg.name}}">
        <Feature name="elidable" value="true"/>
    </Subj> 
    <VP word="manages"/>
    <Obj> 
        <NP string="a team"/>
        <NP preposition="of">
            <Number int="{{msg.teamSize}}" isAlpha="true"/> 
        </NP>
    </Obj>   
</S>

Equivalent parse trees for the lex rules are shown below:

The second noun phrase (NP) in the object of the second rule is actually a prepositional phrase (shown as PP in the image); this is because there is preposition="of" in the opening tag of that NP.

Both lex rules have a new feature: <Feature name="elidable" value="true"/>. Because the feature is added inside the subject (i.e. inside the Subj tabs), it enables the subjects of sentences generated by this rule to be elided in certain contexts.

JSON data values are accessed in the lex rules as before with "{{msg.name}}", "{{msg.job}}", and "{{msg.teamSize}}". The JSON data values are accessed through the keys (“name”, “job” and “teamSize”) for the section of the JSON data that is input to the rules.

Let’s generate sentences by calling the realise function twice, each time with a different lex rule and using the entire “employees” array:

Remember that the realise function’s parameter is a tuple. In this case, the first realise function contains a tuple with the entire array WholeJSON.employees and the value of "classes" specified in the opening tag of the first lex rule, jobClass. The second realise function’s tuple has the same JSON data but the value of "classes" is teamClass from the second lex rule.

There is no subject elision because the subjects are all different!

Now we’ve seen that both lex rules work. Let’s call the realise function with two tuples using the first element in the “employees” array and both lex rules. Now we get subject elision:

The sentences from the two lex rules are joined and the subject of the second has been elided.

Next, we’ll call the realise function with two tuples using each element of the "employees" array in turn and both lex rules. The result is subject elision on all output sentences:

The forAll function goes through the "employees" array and calls realise on each element in the array using both lex rules. The result is four aggregated sentences with subject elision, one for each person’s data.

Finally, we’ll create a user-defined function, describeEmployee, that uses the employee’s name as a subheading and a call to realise with two tuples as before:

#define describeEmployee(employee)

[[employee.name]]
[[ realise((employee,"jobClass"),(employee,"teamClass" )) ]]

We call describeEmployee() in the Main script with the forAll function as follows:

[[forAll(WholeJSON.employees, employee -> describeEmployee(employee) )]]

The result is that each element in the array in turn is input to our user-defined function, describeEmployee:

The effect of subject elision in our examples is to join information about each person in a single sentence, as shown in the image below: