Skip to content

Commit e15bb73

Browse files
aaronjnewmanclaude
andcommitted
Add reviewer recruiting email draft
Template for asking department alumni/faculty to help clean up OCR errors via the corrections Google Sheet. Personalize the opener; everything below "What it is" is reusable. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 parent 82b6b49 commit e15bb73

1 file changed

Lines changed: 93 additions & 0 deletions

File tree

scripts/recruiting_email.md

Lines changed: 93 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,93 @@
1+
# Recruiting email — draft
2+
3+
Personalize the opener for each recipient (and add specific year
4+
suggestions where you can — e.g. "Suzanne, you've already done a pass on
5+
1995–2000, want to keep going from there?"). Everything below the
6+
"What it is" section can stay as-is.
7+
8+
---
9+
10+
**Subject:** Help clean up 50 years of In-House Conference abstracts?
11+
12+
Hi {Name},
13+
14+
For the 50th-anniversary In-House this year I put together a complete
15+
searchable archive of every program from 1975 to 2026 — roughly **2,150
16+
abstracts** across 48 conferences (2020 and 2021 are absent — COVID).
17+
The scans and PDFs you and others rescued have been OCR'd into a
18+
structured database and pushed to GitHub:
19+
20+
<https://github.com/aaronjnewman/inhouse-conference-archive>
21+
22+
The OCR did its best, but the older scans in particular have lots of
23+
small errors — author names mangled (`Hc1,Jaughton` for `McNaughton`),
24+
dropped accents, occasional words swallowed up. I'd like a small group
25+
of folks who know this department to do a careful pass year-by-year and
26+
clean those up.
27+
28+
## What it is
29+
30+
A **Google Sheet, one tab per year**. No GitHub account, no software
31+
to install. Each row is one presentation. You see what the OCR captured
32+
on the left, and there are **yellow columns** on the right where you
33+
type the corrected version.
34+
35+
## What you'd do
36+
37+
1. I'll share the sheet with you. Pick one or more years.
38+
2. On each row: if the title / authors / abstract is wrong, type the
39+
correct version in the yellow column next to it. If it's already
40+
right, leave the yellow columns empty.
41+
3. Set the row's **`status`** dropdown:
42+
- `fixed` — you typed at least one correction
43+
- `checked-ok` — you read it and nothing needed changing
44+
- `needs-help` — something's wrong and you're not sure how to fix it
45+
4. Put your name in the **`reviewer`** column so we can credit you.
46+
5. Each row has a **`source_pdf`** link straight to the program PDF on
47+
GitHub — click it if you want to compare against the original scan.
48+
49+
## A few conventions
50+
51+
- **Author format:** `Lastname, Firstname and Lastname, Firstname`
52+
e.g. `LoLordo, Vincent M. and Jacobs, W. J. and Foree, Donald D.`
53+
This is BibTeX's canonical form. It avoids ambiguity with compound
54+
surnames ("Van der Berg") and initial-only given names ("M.G. Yoon").
55+
- **Don't touch the white columns** — those are the read-only "what the
56+
OCR currently says" record.
57+
- **Don't worry about structural problems** — e.g. one abstract that's
58+
actually two presentations fused together, or an entry that's missing
59+
entirely. Just set `status: needs-help` and explain in `notes`; I'll
60+
fix those by hand.
61+
62+
## Time
63+
64+
Each year has 15–65 entries. A careful pass on one year is roughly
65+
30–60 minutes depending on how bad the OCR was. There's no deadline —
66+
do as much or as little as you want.
67+
68+
## How fixes get applied
69+
70+
Once a week, a script pulls the corrections from the sheet, applies
71+
them to the canonical archive, and regenerates the BibTeX file. So
72+
every Monday the public archive on GitHub reflects everyone's work
73+
from the prior week.
74+
75+
If you're game, just reply and I'll share the sheet. Happy to do a
76+
15-minute screenshare first if you'd like a walkthrough.
77+
78+
Thanks for considering it — this is exactly the kind of thing only
79+
people who've been around the department can do well.
80+
81+
Aaron
82+
83+
---
84+
85+
## After they reply
86+
87+
Things you'll likely need to do once they say yes:
88+
89+
- Share the spreadsheet with their personal Gmail (Editor access).
90+
- Optionally suggest 1–2 specific years they have personal knowledge of
91+
("you were a regular through the late 90s — would you take 1996–2000?").
92+
- Point them at the **status legend** at the top of `scripts/README.md`
93+
if they ask for more detail than the email gives.

0 commit comments

Comments
 (0)