The Decoder project, funded by the European Union's Horizon 2020 program, aims to enhance software development through formal documentation and knowledge extraction using natural language processing. The project encompasses various tools and methodologies to support the entire development lifecycle, including code analysis, documentation generation, and evaluation of external software. Current efforts focus on implementing a persistent knowledge monitor database and refining natural language understanding features while seeking user feedback and collaboration with other projects.

1. DEveloper COmpanion for Documented and annotatEd code Reference
The DECODER project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 824231.
DECODER Project Overview
Virgile Prevosto
with DECODER's partners
Advisory Board Meeting
2019-07-11

2. Outline
Project Objectives
Initial Results
Roadmap and Conclusion
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
2

3. DEveloper COmpanion for Documented on annotatEd
code Reference
Persistent Knowledge Monitor
Database storing all relevant
documents for a software project
documentation (manuals,
comments, bug tracker, ...)
formal specifications
source code
analysis and testing results
and evolutions (e.g. commits)
Tools for feeding the database
Tools for querying the database
Common schema to ease interactions
PKM
Testar
Frama-C
OpenJML
Specification
synthesis
Moskitt
Documentation
generation
NLP
knowledge
formalizer
NLP
knowledge
extractor
Syntactic
analyses
Augmented
IDE
ASFMmodels
inferred
properties
analysis results
infer properties
UM
L
m
odels
Inferredproperties
M
odels, Properties
Code, documentation
raw code information
Advancedqueriesoncode
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
3

4. Encompassing the Whole Development Lifecycle
Requirements
Preliminary Design
Detailed Design
Implementation
Unit Checking
Integration Checking
System Validation
Maintenance
Evolution
NLP
spec generation
Modeling
ASFM
doc generation
IDE queries
invariant
generation
PKM
traceability
doc generation
PKM
Impact analysis
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
4

5. From Informal to Formal Documents
Natural Language Processing
Knowledge extraction from informal documents and correspondance with relevant
code pieces
Knowledge extraction from code and semi-automated documentation generation
Abstract Semi-Formal Models (ASFM)
Graphical language to describe effects of a function on the data structures involved
Semi-automated generation of ASFM diagrams
Animation of the diagrams (graphical debugging)
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
5

6. Use Cases
Evaluation and Improvement of DECODER toolset
Linux Drivers Quickly and accurately assess the quality of an external Linux driver for
inclusion in embedded systems
OpenCV Build a better knowledge of OpenCV API and its usage in some
applications
MyThaiStar UI/UX design and verification
Java Usage of Decoder toolset on selected Open Source Java projects
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
6

7. PKM Meta Model
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
7

8. PKM Server Design
Json as main interchange format
Start working on a Json Schema
Take advantage of existing
proposals:
SARIF
JCDB
Others?
Back-end: Document-oriented DB
MongoDB: licence issues
CouchDB
OrientDB: graph model used by
Testar
Others?
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
8

9. Information Extraction
Dataset Gathering
Collect existing datasets (code and documents) outside Decoder
Start looking at use cases (MyThaiStar and OpenCV)
DeepAPI training corpus for natural language/call sequences correspondance
Initial experiments
Code to NL:
features extraction and grouping tokens
NL to Code:
Consider programming language as a foreign language
Neural Machine Translation
Objective: compute semantic similarity between source code and informal description
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
9

10. UX Workshop
Held at F2F meeting in Valencia on
June, 26th
Prepare design of PKM client(s)
Define core roles
Define usage scenarios
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
10

11. Dissemination Activities
Public website at
https://decoder-project.eu
Present on LinkedIn and Twitter
Decoder poster, presented at OW2 Conf
Contact with other projects
https://openreq.eu/
https://www.chariotproject.eu/
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
11

12. Current Roadmap
Progress according to plan
PKM implementation still the main challenge
Schema design to start as soon as possible
DB engine final choice less stringent
NLP and code processing in good shape
Work on ASFM starting this month
Use cases investigation and methodology discussions ramping up
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
12

13. Open Questions
What should fit in the schema?
Suggestion on a particular document DB implementation?
Projects with which we could interact?
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
13

14. Thank You!
The DECODER project has received funding from the European Union’s Horizon 2020 research and innovation programme under
grant agreement number 824231.
If you need further information, please contact the coordinator: TECHNIKON Forschungs- und Planungsgesellschaft mbH, Burgplatz
3a, 9500 Villach, AUSTRIA, Tel: +43 4242 233 55, Fax: +43 4242 233 55 77, E-Mail: coordination@DECODER.eu
The information in this document is provided “as is”, and no guarantee or warranty is given that the information is fit for any particular purpose. The content of
this document reflects only the author`s view – the European Commission is not responsible for any use that may be made of the information it contains.
The users use the information at their sole risk and liability.
2019 DEveloper COmpanion for Documented and annotatEd code Reference 2019-07-11
14