77 lines
6.2 KiB
Markdown
77 lines
6.2 KiB
Markdown
type:: [[REVIEWS]]
|
|
tags::
|
|
year:: 2023
|
|
venue:: [[SCP]]
|
|
full-title:: An industrial experience report on Model-based, AI-enabled Proposal development for an RFP/RFI due to Science of Computer Programming
|
|
date-start:: [[02-09-2023]] - 16:54
|
|
date-submitted::
|
|
external-links:: [Editorial Manager®](https://www2.cloud.editorialmanager.com/scico/default2.aspx)
|
|
status:: [[DONE]]
|
|
deadline-submission:: [[21-09-2023]]
|
|
file:: 
|
|
|
|
- [[Highlights]]
|
|
- #+BEGIN_IMPORTANT
|
|
((651085c3-3715-404e-9100-5c24ebef1976))
|
|
((651085d8-2efb-46fc-91c5-aae0fef6af0c))
|
|
((65108a23-1148-47c8-8277-20e98018b7bd))
|
|
((65117731-fe4f-4e22-ba61-3e70c66f8ed1))
|
|
((65117bfa-52f9-44fd-8933-ef4941495454))
|
|
#+END_IMPORTANT
|
|
- ((65108970-9dce-43c0-aa87-60270a0002af))
|
|
- Any document management system can help?
|
|
- Fig. 1 is representing situations that should not occur if document management systems are used! Very unfortunate and basic situations are shown. Nowadays they should no longer happen!
|
|
- ((651089e7-5fa9-44e7-a613-57e128bd7a56))
|
|
- Can this be generalized as typical scenario?
|
|
- ((65108a8e-fcee-4d96-b310-5f4801a6f54a))
|
|
- What is it?
|
|
- ((65108acf-c3a9-421a-b8ad-9112ec9489b0))
|
|
- The whole paragraph should benefit the description of a concrete and illustrative example, especially to show the quality of the generated proposal document out of the given RFx as input.
|
|
- ((65108eb3-048e-470a-9f8b-1dc53786e00f))
|
|
- Proposal documents should be creative artifacts... how can this be automatically generated????
|
|
- ((65108feb-8432-4876-ba35-3d1b9d57e49e))
|
|
- What does it mean. There are many sentences like this one.
|
|
- ((65109043-3284-4742-9433-d7967d3ebc4e))
|
|
- How knowledge is stored and indexed? what are the facets that have been defined to index documents?
|
|
- ((6510a332-3685-46b1-a58b-a4bb49e0d8ff))
|
|
- The data of the "encoding" are not clear
|
|
- ((6510a48e-5aad-4067-8fac-f0a6449b742e))
|
|
- I guess the recommender produces a ranked list of recommendations. Is the modeltotext phase applied only to the first one?
|
|
- ((6510a554-e14d-4ec9-a62a-ea44c185aa75))
|
|
- You should elaborate more on the knowledge base that has been curated to let the recommender perform as expected. Being it data-driven, the creation of high-quality knowledge base is crucial!
|
|
- ((6510a5df-3cd0-43b1-a38b-71343422bf55))
|
|
- Are you sure it always properly work?
|
|
- ((6510a646-603b-439c-96ea-a77533810599))
|
|
- The algorithm should be better described.
|
|
- ((6511750e-a50b-40d0-98ba-1ad38c3ceb0f))
|
|
- It's not clear the goal of this section.
|
|
- ((65117555-0030-4899-9f5e-5e6704321216))
|
|
- It's a very small dataset to actually support the creative generation of documents satisfting/meeting the requirements given in RFx.
|
|
- ((651175a7-aece-49f3-b6e9-ad473abae0ee))
|
|
- It's an organized visualization/management of documents, isn't it?
|
|
- ((6511769b-73ae-4190-8878-a335a888893c))
|
|
- Look very application specific..... It seems that the system can increase the reusability of stored documents by properly storing and indexing them.
|
|
- ((65117c30-8209-4797-a457-ff9e5bfe1ca9))
|
|
- Only the metamodel concept is essentially used, right?
|
|
- ((6512e053-caf4-4086-aa3c-333adb8e98c6))
|
|
- This needs to be demonstrated!!!
|
|
- To what extent the document structure and thus the corresponding elements reflect the semantical inforamtion of each document that is supposed to answer very specific requests! The fact that the metamodel is domain-agnostic makes this concerns even more critical!
|
|
- The following is related to this comment:
|
|
- ((6512e165-d6e3-43b1-b085-a8e92e2c3555))
|
|
- ((6512e184-4384-4e04-b59f-26e64d693569))
|
|
- The process seems to be very strict to me and it is not clear how to consider the semantics of the document!
|
|
- ((6512e1da-5919-417f-a514-0227aec42646))
|
|
- This is very strange and in contrast with the fact that the parser of the document is aware of the different constituting elements of the elements and that need to be represented in the model being created and conforming to the proposal metamodel!
|
|
- ((6512e3c3-bbe3-420c-adce-a1e072e89e8c))
|
|
- Indeed!
|
|
-
|
|
- [[Comments]]
|
|
- The paper introduces a model-based approach designed to facilitate the creation of documents in response to Requests for Proposals (RFP) and Requests for Information (RFI). This methodology relies on the availability of a knowledge base, which is constructed by parsing existing RFx documents and representing them as models conforming to a predefined proposal metamodel. When a new proposal needs to be generated, the user specifies a set of queries, which are then extracted and utilized as queries to ultimately produce a proposal document in alignment with the user's initial requirements. Through experimental evaluation, it was found that the proposed tool can automatically generate approximately 70% of proposal documents effectively.
|
|
- Overall the paper is about an interesting topic. The revised submission tackles most of the concerns raised during the first round of review. However, I still have some doubts about the effectiveness of the approach, primarily related to two key issues:
|
|
- The approach heavily relies on the structural characteristics of the document, and it appears that the semantics of the content are only considered to a limited extent. Crafting a proposal that responds to user requests across potentially any domain is a creative process that can only be automated to a partial degree and heavily relies on the quality and diversity of the training data.
|
|
- The experiments indicate that 70% of the proposal documents can be automatically generated using the proposed tool. What remains unclear is what happens with the remaining 30% and the effort required to create it. This aspect needs to be addressed consistently with the 70% that can be automated. Users must have a clear understanding of how to extend the automated portion to ultimately obtain the final document.
|
|
- Addressing these two concerns is crucial to make the paper more convincing and to present the approach as a genuine success story in the application of Model-Driven Engineering (MDE) to a very specific domain.
|
|
- [[REVIEWS/Notes]]
|
|
-
|
|
-
|
|
- |