The show-and-tell-play-in method (S&T-play-in) uses online parsing and the state of the current parse to
interpret the interaction and integrate it into the scenario-based requirement; i.e., into the LSC that is being
constructed on the fly.
An interaction can be interpreted in multiple ways. When an object is selected (from the model or the
GUI), either its name or the operation performed on it (e.g., clicked, or turned) may be integrated into the
sentence. When an object property is selected, it may be a reference to the property name or to the property
value. The parsing is performed bottom-up using an active chart parser similar to that of Kay (1986) and
adapted for online parsing as in Jurafsky and Martin (2009), Figure 1a shows the system architecture and
Figure 2 provides details of the algorithm.
In each requirement being entered, the indexes represent the locations between the words (as in 0 when 1
the 2 user 3). An edge represents a grammar rule and the progress made in recognizing it. We use the common
dotted rule, where a dot ('') within the right-hand side of the edge indicates the progress made in recognizing