Reviewer has chosen to be Anonymous

Summary of paper in a few sentences:

The paper proposes a task recommendation system to guide employers in matching candidates (internal or otherwise) to open positions. The method takes as input candidates’ resumes and information on the skillset required for the job position, and outputs a matching-based classification. The method is validated on an example dataset.

Reasons to accept:

None

Reasons to reject:

This is a list of reasons to reject. In “further comments” I will elaborate on each of the items.
> The paper is not correctly placed in the literature
> The scientific contribution is not clear
> The proposed method does not seem to apply to the context that the authors want to study
> The quality of the presentation is unacceptable

Nanopublication comments:

Further comments:

> The paper is not correctly placed in the literature
There are two major issues with the placement in the existing literature. On the one hand, the literature on job allocation and resume application from HRM and business communication is almost completely ignored. On the other hand, the few in-text citations featured in the paper do not relate to the text they are attached to. Examples include:
Page 2, “Screening process”, first paragraph [typos are in the original text]: “Many screening methods are used for recruitment like aptitude round, group discussion, Personal interview etc. but initial screening is usually based on resumes.(Nunley, 2016)”. While this claim is compatible with common sense, it cannot by attributed to Nunley et al. (2016), whose study only relates to this claim to the extent that they compare resumes in their study.
A couple of sentences later, the text reads [again, the typo is in the original text]: “A good resume is meant to provide a complete picture about the potentials of candidate and reflects whether the applicant is worth interviewing.(Derous, 2017)”. Here, too, the only connection between the sentence and the study by Derous et al. (2017) is that Derous et al. study *something* about resumes (specifically, ethnic discrimination in resume screening).
One last example from the top of page 3: “Recommending can help users to focus on the relevant news instead of getting confused by the bombardment of too many news documents. (Miller, 2003) (Melville, 2009)”. The paper by Melville et al. (2009), which is about sentiment analysis on weblogs, has nothing to do with what is being argued in the text.

> The scientific contribution is not clear
The paper does not explain what the novelty of the study is, and thus it fails to identify the niche it addresses.

> The proposed method does not seem to apply to the context that the authors want to study
I have two fundamental questions that potentially undermine the premise of this study.
1) Semantic similarity is at the core of the proposed task recommendation system, and is computed by “grouping the terms into synsets as provided by WordNet”. While the identification of sematic similarity works accurately with common, ‘general’ vocabulary, it is a much harder task when dealing with specialized jargon and technical vocabulary. Job applications, CVs, and job descriptions typically sport specialized jargon, and I can imagine that such jargon is used to identify the skills that the potential employer is trying to match with the job requirements. Hence my question: are job applications a plausible setting to study the task recommendation system?
2) Am I right to assume that the recommender system can very easily be tricked by purposively crafted resumes?

> The quality of the presentation is unacceptable
The language is poor to the point that some sentences are incomprehensible. The diagrams are all either non informative, not correctly displayed on the page, ruined by formatting errors, or a combination of the above. Citations and references are not reported in any standard (or at least consistent) layout.

Reviewer has chosen to be Anonymous

Summary of paper in a few sentences:

The manuscript describes a method for keyword-based matching of tasks and resumes. Competencies of workers are extracted out of their resumes on a word occurrence count basis, then used for clustering the resumes. This classification has two implications: (1) tasks can be assigned to the most relevant cluster of resumes, (2) workers can be advised to update their skillset based on competencies available in their cluster.

Reasons to accept:

The method laid out in this paper might be novel and innovative. It might also concern a problem that is pressing and relevant in very large companies with a structured, yet dynamic task flow.

Reasons to reject:

Most importantly, it is unclear to the reader why the problem that is addressed in this manuscript (assigning tasks to workers) would be an important one to solve. Perhaps the authors could elaborate on the contexts in which they reckon this tool would be useful (types of companies, types of tasks). In general the method strikes me as rather imprecise to assign specific tasks to laborers in the broadest sense of the word.
Besides the uncertainty about its application, the scientific relevance is not clearly stated. There is some literature cited here and there, but it is not obvious whether and, if so, where the gap in the literature exactly is.
The manuscript contains a range of grammar, style, and simple typing errors that should be fixed before publication.
The manuscript does not use a consistent citation style.
The tables and figures in the manuscript are messy and inconsistent.
The structure of the text and headings is messy, which makes it difficult for the reader to grasp what the authors want to convey, and why.

Nanopublication comments:

Further comments: