CIV 1320 Indoor Air Quality

September 11, 2024

Reading Assignment

The objective of this assignment is to develop skills in reading and extracting information from articles on indoor air quality in the literature as well as some experience with using and evaluating machine learning/natural language models for indoor air research. Your overall goal is to use AI tools to develop a literature review and then critique it. You don’t have to strictly  follow the approach below, but pay attention to the evaluation criteria. In particular, certain AI  tools are described below. They continually update their sign-in requirements and availability and so you can use alternative tools to achieve the same aims.

Step 1. Pick your topic.  It can be anything related to indoor air quality.  Note that this will be a little iterative with later steps, but your goal at this point is to understand the landscape of the available literature on the topic. Try different variations of your topic in a conventional scientific literature search engine (eg, Science Citation Index or Google Scholar). Document this literature search and how you refined it. Identify and note the important articles on the subject.

Step 2. Use Elicit (https://elicit.com/) or similar (you can also use the old version of Elicit (https://old.elicit.org/)), to find top articles on your chosen subject. Sometimes Elicit responds better to questions than to topics and so you might need to play around with different variants of the input. Once you are happy with the summary, note the top four articles and copy the summary. Document your question/topic refinement.

Step 3. Paste your copied summary into a generative AI tool (eg, Chat GPT or similar). Note that the tool that is recommended for the UofT community is Microsoft Copilot (https://teaching.utoronto.ca/tool-guides/microsoft-copilot/). Tell it to expand the text/make it longer (you might have to do this several times) until you have a document that is about two pages long. Sometimes the expansion doesn’t continue past a point and so don’t worry if you don’t get fully to two pages.

Step 4.  Get the four articles found in Step 2 (or the four most important articles for the generated summary).  Read them with enough depth that you can comment on the generated text.

Step 5. Address the following questions in a cohesive narrative that is <3 pages in length (12 pt Times New Roman single spaced, 1” margins,

1.   Are the four articles, the ones you would select on the topic?  Why or why not?  Are there any articles that shouldn’t be in the summary?  If yes, can you refine the question to get a different article? If you could add one article to the review, which would you pick?  What would it add to the summary.

2.   Address the question of accuracy. Does the  summary text contain any inaccuracies or critical omissions (an article says something essential that the summary misses) from the four articles?  Detail these inaccuracies/omissions and your level of concern about them,

3.   Address the question of depth.  Does the summary text do a good job summarizing the nuance and detail of the included papers?  Justify your answer with specific examples.

4.   Find a review paper (including keynote and overview articles) in the literature  that addresses your topic (or as close as you can get to it) that is not one of the five articles in Step 1.  Consider the review article to the AI-generated article from the perspective of meeting their stated objectives and not directly to each other. What are the strengths and weaknesses of each of these approaches. What would a procedure

5.   It is important to reflect on who/what we cite in scientific research because of the potential for bias, among other concerns. Do all of the articles in Step 1 come from high-quality peer-reviewed journals? Do they represent reasonable diversity (as you define it) in terms of authors, time periods, research groups, institutions, geographic locations, etc.? Justify your answer.  Pick one area of diversity that is not well-represented and try and locate an article that satisfies this area. Would inclusion of this article improve a literature review? Justify your answer.

Deliverable and evaluation (all are single-spaced, 1” margins, 12 pt Times New Roman)

Steps 1 & 2 Documentation.  < 1 page                                                                                     10%

Logical process for refinement

Documentation clear and easy to follow

Research question/topic appropriate for review (eg, not overly broad or narrow)

Step 3 AI-generated summary  ~ 2 pages                                                                                10%

Completion according to criteria

Step 4 - 5 Narrative                                                                                                                   80%

Writing quality and clarity (20%)

<3 pages in length, 12 pt Times New Roman single spaced, 1” margins Completeness of response to questions that reflect depth and accuracy (60%)   Additional (5.1, 5.4, 5.5) articles quality and justification (20%)

Include references to cited works in the Reference list. Do not use any blank lines between references.. The references should be in the order they first appear in the text and follow any consistent and complete format (include titles).  Reference do not count towards your page limit

1. ASTM International (2012) Designation: D6245-12 Standard Guide for Using Indoor Carbon Dioxide Concentrations to Evaluate Indoor Air Quality and Ventilation. ASTM Interntional.

2. Hospodsky D, Qian J, Nazaroff WW et al (2012) Human Occupancy as a Source of Indoor

Airborne Bacteria. PLoS ONE 7(4): e34867. doi:10.1371/journal.pone.0034867

3. Obbard JP, Fang LS (2003) Airborne Concentrations of Bacteria in a Hospital Environment in Singapore. Water, Air and Soil Pollution, 144(1), 333-341.

4. Santin OG, Itard L, Visscher H (2009) The effect of occupancy and building characteristics on energy use for space and water heating in Dutch residential stock. Energy and Buildings, 41(11), 1223-1232.