Output

Having completed the population of all internal data structures, the system is capable of outputting the following information:

• Similarity values ( $0 \leq n \leq 1$)
• Smoothed similarity values ( $0 \leq n \leq 1$)
• Detected topic breaks (integers representing pseudosentence numbers)
• Human-annotated topic breaks
• Dialogue in XML format, with pseudosentence numbers marked, and optionally with automatic or hand-transcribed topic breaks marked

Each data type (except for the XML format) is output in a tab-separated flat-file, with a comment (signified by ``#'') naming the columns. For example, a smoothed-similarity-values output begins:

#Pseudosentence SmoothSimilarity
5               0.622067
6               0.634649
7               0.615261
8               0.566583
9               0.501763

...