Visual Narrator

Tells Your User Story Graphically

This program turns user stories into a conceptual model containing entities and relationships.

Input

• Text file (.txt, .csv, etc.) containing one user story per line

Output

• Report of user story parsing, and conceptual model creation
• Manchester Ontology (.omn) describing the conceptual model
• (Optional) Prolog (.pl) arguments
• (Optional) Statistics about the user stories
• Returns the mined user stories, ontology, prolog and matrix of weights, which can be used by other tools

Dependencies

The main dependency for the program is its Natural Language Processor (NLP) spaCy. To run the program, you need:

• Python >= 3.4
• spaCy >= 1.1.0 (currently under development using v1.6.0)
• NumPy >= 1.7
• Pandas >= 0.17.1
• Jinja2 >= 2.8

Running the Project

Running the program can only be done from the command line. With the program main directory as current directory, run the program by executing:

python run.py <INPUT FILE> [<arguments>]

Arguments

The most important arguments is INPUT FILE to specify the location of the text input file. The table below provides an overview of the currently implemented arguments.

Positional arguments

Argument	Required?	Description
INPUT FILE	Yes	Specify the file name of the User Story input file

Optional arguments

General

Argument	Description
-h, --help	Show a help message and exit
-n SYSTEM_NAME, --name SYSTEM_NAME	Specify a name for your system
-u, --print_us	Print additional information per User Story
-o, --print_ont	Print the output ontology in the terminal
--prolog	Output prolog arguments to a .pl file. Combine with --link to reason about user stories
--version	Display the program's version number and exit

Statistics

Argument	Description
-s, --statistics	Show statistics for the User Story set and output these in .csv files

Ontology generation tuning

Argument	Description	Type	Default
-p, --per_role	Create an additional conceptual model per role	N/A
-l, --link	Link all ontology classes to their respective User Story for usage in the set analysis	N/A
-t THRESHOLD	Set the threshold for the selected classes	FLOAT	1.0
-b BASE_WEIGHT	Set the base weight	INT	1
-wfr WEIGHT_FUNC_ROLE	Weight of functional role	FLOAT	1.0
-wmo WEIGHT_MAIN_OBJ	Weight of main object	FLOAT	1.0
-wffm WEIGHT_FF_MEANS	Weight of noun in free form means	FLOAT	0.7
-wffe WEIGHT_FF_ENDS	Weight of noun in free form ends	FLOAT	0.5
-wcompound WEIGHT_COMPOUND	Weight of nouns in compound compared to head	FLOAT	0.66

Example usage

python run.py example_stories.txt -n "TicketSystem" -u

Conceptual Model

The classes in the program are based on the following conceptual model:

The Reader starts by reading the input file line by line and generates a list of sentences. These sentences are then enriched using Natural Language Processing, adding Part-of-Speech tags, dependencies, named entity recognition, etc. Subsequently, the StoryMiner uses these enriched sentences to create UserStory objects. The User Story objects contain all the information that could be mined from the sentence. These are then used to attach weight to each term in the User Story, creating Term-by-US Matrix in the Matrix class. The Constructor then constructs patterns out of each user story, using the Term-by-US Matrix to attach a weight to each token. The Constructor forms a model for an ontology, which is then used by the Generator to generate a Manchester Ontology file (.omn) and optionally a Prolog file (.pl). Finally, these files are printed to an actual file by the Writer in the '/ontologies' and '/prolog' folders respectively.