# SubResearch

Kho nay duoc to chuc cho workflow rat don gian:

1. Bo paper moi vao `inbox/`
2. Codex doc ky paper ngay tu luot dau tien
3. Codex chon `primary group`
4. Codex tao `paper note` chuan
5. Codex trich cac `concept notes` can tai su dung
6. Codex cap nhat `synthesis note` cho nhom lien quan
7. Codex cap nhat `index.html` o root de co trang tra cuu nhanh

## Cau truc toi gian

```text
SubResearch/
|- paper-drafts/
|- inbox/
|- papers/
|  |- submodular/
|  |- k-submodular/
|  |- dr-submodular/
|  `- mixed/
|- concepts/
|- syntheses/
|  |- submodular.md
|  |- k-submodular.md
|  |- dr-submodular.md
|  `- mixed.md
|- templates/
`- scripts/
```

## Vi sao cau truc nay gon hon

Khong co gang chia nho san theo qua nhieu tieu muc nhu `algorithms/`, `constraints/`, `guarantees/`, vi nhieu paper giao thoa nhau va rat kho ep vao mot o duy nhat.

Thay vao do:

- `paper-drafts/` luu cac ban thao paper ma ban dang viet
- `papers/` luu paper goc va note gan paper
- `concepts/` luu dinh nghia, ky hieu, lemma, relation co the tai dung
- `syntheses/` luu tong hop theo nhom lon
- `tags` trong metadata dung de bo sung phan loai mem

## Nhom chinh

Paper moi chi can chon `primary group`:

- `submodular`
- `k-submodular`
- `dr-submodular`
- `mixed`

Neu paper giao thoa, khong can co phan dinh qua cung. Chi can:

- chon mot `primary group` de dat vao dung cho
- them `secondary tags` trong `metadata.md`

`mixed/` dung cho nhung paper that su nam giua nhieu nhom hoac kho chot mot nhom chinh.

## Moi paper co duoc doc 1 lan khong?

Co. Workflow mac dinh la:

1. Codex doc paper 1 luot dau, nhung doc ky
2. Tao mot `canonical paper note` muc tieu la du chi tiet de khong phai doc lai toan bo PDF nhieu lan
3. Tu note do, trich ra:
   - concept nao nen dua vao `concepts/`
   - thong tin nao nen cap nhat vao `syntheses/`

Khong can moi lan can tra cuu lai mo PDF tu dau. Ve sau chi quay lai paper goc khi:

- can kiem tra chi tiet proof
- can doc sau hon mot lemma
- can so sanh voi paper moi

Khi doc paper lan dau, nen co gang ghi luon:

- bai toan va setting chinh xac
- theorem / guarantee chinh
- y tuong thuat toan
- y tuong phan tich / chung minh
- cac lemma quan trong
- assumptions, caveats, va diem de nham

## Y nghia tung thu muc

### `inbox/`

Noi de tha paper moi vao truoc khi xu ly. Co the bo vao:

- file `.pdf`
- source `.tex`
- zip source
- appendix / slides

Day la diem vao duy nhat cho paper moi.

### `paper-drafts/`

Noi de luu cac ban thao paper ma ban dang viet.

Thu muc nay khong thay the `inbox/` hay `papers/`:

- `inbox/` van la intake raw cho literature ben ngoai
- `paper-drafts/` la khu vuc viet / soan thao
- `papers/` van la canonical archive cua literature da doc

### `papers/`

Sau khi Codex classify, moi paper se duoc dua vao:

```text
papers/<primary-group>/YYYY-firstauthor-short-title/
```

Ben trong folder:

- `paper.pdf`
- `source/`
- `metadata.md`
- `summary.md`

### `concepts/`

Noi luu cac note tai su dung duoc, khong gan rieng mot paper:

- definitions
- notation
- standard lemmas
- relations between models
- recurring techniques

Khong can chia nho san. Moi concept la mot file.

### `syntheses/`

Moi file la mot ban tong hop song song theo nhom lon:

- [submodular.md](C:\Users\Admin\Documents\GitHub\SubResearch\syntheses\submodular.md)
- [k-submodular.md](C:\Users\Admin\Documents\GitHub\SubResearch\syntheses\k-submodular.md)
- [dr-submodular.md](C:\Users\Admin\Documents\GitHub\SubResearch\syntheses\dr-submodular.md)
- [mixed.md](C:\Users\Admin\Documents\GitHub\SubResearch\syntheses\mixed.md)

Muc tieu la tra loi nhanh:

- nhom nay hien co nhung huong nao?
- paper nao quan trong?
- techniques nao lap lai?
- co khoang trong nao trong van hoc?

## Trang index nhanh

Tai root co file `index.html`.

Day la trang tra cuu nhanh de:

- nhay thang vao paper canonical
- mo `summary.md`, `metadata.md`, va `paper.pdf`
- di nhanh sang `concepts/`, `syntheses/`, va `inbox/intake-index.md`

Quy uoc tu nay:

- moi khi them paper canonical moi
- moi khi doi status quan trong
- moi khi co concept / synthesis moi that su can hien len mat ngoai

thi cap nhat `index.html` cung luc.

## Metadata toi thieu

Trong `metadata.md`, nen co:

- title
- authors
- year
- venue / arXiv
- primary group
- secondary tags
- problem setting
- main result
- key techniques
- status

Vi du tags:

```text
#submodular #matroid #nonmonotone
#dr-submodular #continuous #lattice
#k-submodular #bisubmodular #approximation
```

## Workflow xu ly paper moi

1. Tha paper vao `inbox/`
2. Bao Codex doc file do
3. Codex chon `primary group`
4. Codex tao folder trong `papers/<group>/...`
5. Codex viet `metadata.md` va `summary.md` o muc du sau nhat co the trong lan doc do
6. Codex trich concept can thiet vao `concepts/`
7. Codex cap nhat file tong hop trong `syntheses/`
8. Codex cap nhat `index.html`

## Nguyen tac chinh

1. Mot paper chi can mot `paper note` chuan.
2. Khong co gang phan loai qua min.
3. Dung `primary group + tags` thay vi cay taxonomy qua sau.
4. Cac khai niem co ban de trong `concepts/`, khong lap lai nhieu noi.
5. Mac dinh la doc sau ngay tu lan dau, khong chi lam first-pass nhe.