Characterizing Text Revisions to Better Support Collaborative
Despite advancement in collaborative writing tools, the track changes capability in modern editors remains limited to highlighting syntactic changes, with authors still required to manually read through each of the revisions. We envision a collaborative authoring system where an author could acc...
Saved in:
| Main Author: | |
|---|---|
| Format: | Proceeding |
| Language: | English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/41196/1/Characterizing%20Text%20Revisions%20to%20Better%20Support%20Collaborative.pdf http://ir.unimas.my/id/eprint/41196/ https://ieeexplore.ieee.org/document/10007395 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Despite advancement in collaborative writing tools,
the track changes capability in modern editors remains limited
to highlighting syntactic changes, with authors still required to
manually read through each of the revisions. We envision a
collaborative authoring system where an author could accept all
minor edits first and then focus on the substantial changes. To
support this, we define the task of significant revision
identification as the task of identifying the revisions between two
versions of a text according to one of four categories, i.e. formal,
meaning preserving, micro- and macro-structure. Micro-
structure change corresponds to minor meaning change while
macro-structure change corresponds to major meaning change.
Our main contribution is to define a computational approach to
this task, by framing the task as bi-directional entailment
between the original and revised sentences. An existing
recognition of textual entailment (RTE) system is applied to
evaluate whether the revised texts entails. We evaluate the
approach through a novel corpus consisting of multiple versions
of drafts of academic papers written by multiple authors, which
were annotated with the four revision types by both authors and
non-authors of the papers. The proposed bi-directional textual
entailment approach performs better than baseline edit distance
approaches, which is similar to the current track changes
capability built into most word processors. |
|---|