Please complete this review form offline and enter into appropriate fields on the Web form. Give reasons that support your answers rather than simply checking boxes.
Papers should address and specify an important problem related to human-level intelligence or complex cognition. Does the submission address such a problem and state it clearly?
Authors should specify their main theoretical tenets, discuss their novelty, and distinguish them from implementation details, if any. Does the paper make clear theoretical statements about representation and/or processing?
Papers should make explicit claims about the capabilities or behavior of the approach. These can take many forms, but they should be stated unambiguously in language accessible to readers. Does the submission make explicit claims?
Authors should present convincing evidence that supports their claims. Such support can take different forms, but should lead a reasonable person to conclude the claims are true or, for negative results, untrue. Does the paper present such convincing evidence?
Submissions should be organized and written in ways that convey their ideas effectively, enough so that informed readers can understand the main contributions and reconstruct the results if desired. Is the paper effective along these lines?
Papers should review prior work that motivated the authors' effort, as well as compare and contrast their approach with other research on similar problems. Do the authors discuss related work adequately?
Based on the above comments, should the paper be accepted for publication? If you favor conditional acceptance, please itemize the changes necessary for publication. These should involve revisions that the authors can plausibly implement in three weeks.
Do you have other comments that support your evaluation or detailed suggestions for improving the paper, whether or not you recommend its acceptance?