From chi2003-shorttalks--at--acm.org Fri Jan 31 13:15:02 2003 Date: Fri, 31 Jan 2003 08:55:53 -0500 From: chi2003-shorttalks--at--acm.org To: ping--at--zesty.ca Cc: chi2003-shorttalks--at--acm.org Subject: CHI 2003 STIP notification - #470 Dear Ka-Ping Yee - Thank you for submitting to the CHI 2003 short talks and interactive posters track. We are sorry to inform you that your paper 470 - User-Directed Screen Reading for Context Menus on Freeform Text has not been accepted to CHI 2003 STIP. We received 324 submissions this year, of which 123 (38%) were accepted. This mail contains the meta-review and reviews for your paper, which we hope you will find of value. Thank you for your participation, and we hope to see you at CHI in Fort Lauderdale, Florida, April 5-10, 2003. Sincerely, Staffan Björk and Jan Gulliksen, CHI 2003 STIP chi2003-shorttalks--at--acm.org ---------------- CHI 2003 STIP paper 470, review 5 ---------------- Title: User-Directed Screen Reading for Context Menus on Freeform Text Reviewer AC Overall rating 3 (scale is 1..5; 5 is best) Reviewer expertise 3 (scale is 1..4; 3 = "Knowledgeable") 1. Expertise 3 (Knowledgeable) 2. The Meta-Review I'm actually more positive about this paper than the reviewers are, but I could only recommend accepting this paper if there was room in the conference. The reviewers do mention issues in the paper that do require attention, but for a short talk, I think it is fine to present a novel concept even without thorough testing of the idea. In this sense, I agree with Reviewer 3's perspective that this is an appropriate submission for a Short Talk. I do agree with the reviewers, especially Reviewer 2, who would like to see more specific description of the mechanisms of how to select an appropriate range of text to get context and how to resolve potential problems (such as more than one date or a range of text that has more than one interpretation or conflicting inferences). Developing robust ways of handling this would appear to be necessary to make this concept useful (and also appears to be a very hard problem). So, as the reviewers point out, these issues make it a flawed submission. But, for a Short Talk, I look for a nugget of a novel concept that is worth sharing with the community, even if the concept is not completely worked out or tested. Thus, I could only recommend accepting this paper if there is room in the conference. Regardless, I would strongly encourage the authors in pursuing this line of research, and learning from the comments raised in the reviews. 3. Acceptance category Either one ---------------- CHI 2003 STIP paper 470, review 1 ---------------- Title: User-Directed Screen Reading for Context Menus on Freeform Text Reviewer reviewer Overall rating 1 (scale is 1..5; 5 is best) Reviewer expertise 3 (scale is 1..4; 3 = "Knowledgeable") 1. Rating 1 (Definite reject: I would argue strongly for rejecting this paper.) 2. Expertise 3 (Knowledgeable) 3. Contribution to HCI This paper describes a system that allows users to select text in an application and take some action with it such as create a calendar entry. 4. The Review This paper is unfortunately not ready for publication. I believe it would require serious editing and polishing before that would be the case. Since short papers need to be camera ready, I gave this paper a 1 rather than a 2. While there is clearly an issue of "english as a second language", the problem with the writing in the paper goes beyond the language issue. It is unclear, informal and not strong enough to be viewed as a scientific publication. In addition to the problem with language and the style of writing, the underlying interaction between the user and the technology is not clearly explained. In reading between the lines, and reading the paper twice, I was eventually able to define what I believed the interaction to be. However, it was not clear to me that the user would fine value in what the system was doing. The paper did nothing to highlight the value of the system findings, nor were any trials conducted with users. There is no discussion of system error rates (how often did it gather the wrong data) or processing latencies. Does "user directed screen reading" mean that the system takes the delimited text (as defined in the paper)and starts to search it for certain string patterns, or does it actually start reading the text outloud as commercial screen readers do? I assumed the former, however it was not clear. Also it was not clear what the authors meant by "visual proximity" as in the sentence "Since visual proximity does not always correspond to textual proximity...". What was visual about using Alt plus right mouse button down? There is no gaze involved. 5. Additional Comments 6. Acceptance category Interactive Poster ---------------- CHI 2003 STIP paper 470, review 2 ---------------- Title: User-Directed Screen Reading for Context Menus on Freeform Text Reviewer reviewer Overall rating 2 (scale is 1..5; 5 is best) Reviewer expertise 3 (scale is 1..4; 3 = "Knowledgeable") 1. Rating 2 (Probably reject: I would argue for rejecting this paper.) 2. Expertise 3 (Knowledgeable) 3. Contribution to HCI This paper describes an approach to seamlessly extract semantic information from plain text under and around a user^Òs cursor. The example case used in the paper is that of extracting (calendar) dates in a number of formats and enabling these to be included as entries into the user^Òs calendar. This approach is an extension to the use of context-based menus, and is similar to the approaches included in the paper^Òs references, where the underlying text is interpreted and options provided. The main extensions are the more advanced pattern matching (semantic interpretation) system employed and the extended selection of text used to extract the information, both of which offer little contribution to the field of HCI. 4. The Review This paper presents a high level view of extended functionality for text based interfaces. The paper focuses on architectural aspects such defining that the text is read by the window manager, but does not discuss the (difficult) aspect of semantic interpretation of the text selected, or how the existing approach delivers a framework for further ^Ótext-helpers^Ô to be constructed. Investigation is required to address the issue of selecting text in cases where a number of dates are included, or when the line length is large (in this case selecting 3 lines of text may include an entire paragraph, resulting in possibly a large number of results, especially if a number of ^Ñtext helpers^Ò are used). 5. Additional Comments This work would benefit form further research, including the identification of a taxonomy of information types (not just dates) that can be extracted from text and other contextual information. 6. Acceptance category Interactive Poster ---------------- CHI 2003 STIP paper 470, review 3 ---------------- Title: User-Directed Screen Reading for Context Menus on Freeform Text Reviewer reviewer Overall rating 4 (scale is 1..5; 5 is best) Reviewer expertise 3 (scale is 1..4; 3 = "Knowledgeable") 1. Rating 4 (Probably accept: I would argue for accepting this paper.) 2. Expertise 3 (Knowledgeable) 3. Contribution to HCI Applies screen reading technology such as that used to adapt interfaces to the blind, to context inference. This allows development of general techniques to infer context and construct context menus from arbitrary text. 4. The Review I found this to be an interesting and well-presented work. I particularly like the emphasis on generalizability. The choice of calendar scheduling for proof of concept is both non-trivial and useful. My only reservation is the question of originality. I do not know of any prior work that specifically covers this topic, but I am not well read in the area of text in user interfaces. It seems to me that this paper is just what a short paper should be, an interesting proposed toolset with a clearly defined purpose and evidence that the approach is workable. 5. Additional Comments 6. Acceptance category Short Talk ---------------- CHI 2003 STIP paper 470, review 4 ---------------- Title: User-Directed Screen Reading for Context Menus on Freeform Text Reviewer reviewer Overall rating 1 (scale is 1..5; 5 is best) Reviewer expertise 3 (scale is 1..4; 3 = "Knowledgeable") 1. Rating 1 (Definite reject: I would argue strongly for rejecting this paper.) 2. Expertise 3 (Knowledgeable) 3. Contribution to HCI This paper proposes a method for making a tool or tools to approve the efficiency of user interaction with GUIs containing text. Such tools could indeed be useful. 4. The Review This paper proposes a method of interaction, and says that a proof-of-concept was built, but offers no test results to show how well this approach works. This is a proposal, not a result. This could be a valuable piece of work when more is done. 5. Additional Comments 6. Acceptance category Short Talk