# BBLAB Bookclub 01:

## 科技論文英文寫作 Chapter 11

導讀者：Chuck

## READING

*每次由導讀者講述對本章節之理解與執得討論之處 10min*

## SHARING

重點分享
- 以辯證邏輯為基礎
- 儘早動手
- 邊工作就要邊想報表怎麼呈現，一邊釐清整個架構
- 有時候方法 結果 討論的切分 ，千萬不要用同樣的句子
- 對結果要說到解釋
- 從多看模仿開始

## CREATING

*每次由導讀者出題，四個題目取自本章節之重點，參與者盡量在當天/隔天內完成*

#### 1) 短句改寫 少就是多 (欣)
> 找一篇科學文章，將口語部分去除並精煉語句，轉換為論文寫作之文字

**原始:**

:::info
Hyperspectral Imaging is a new analytical technique based on spectroscopy. It collects hundreds of images at different wavelengths for the same spatial area. While the human eye has only three color receptors in the blue, green and red, hyperspectral imaging measures the continuous spectrum of the light for each pixel of the scene with fine wavelength resolution, not only in the visible but also in the near-infrared. The collected data form a so-called hyperspectral cube, in which two dimensions represent the spatial extent of the scene and the third its spectral content.
:::

**修改:**
Hyperspectral image acquires the continuous spectrum image in specific spatial area. The collected data, which we call hyperspectral cube not only including spatial information, but also visible and near-infrared bands image information.

**修改邏輯：**
1. 科普文章須對專有名詞說明，但對於對應領域的研究人員，說明可見光包含那些顏色就會顯得多餘。
2. 本段內容重點在說明hyperspectral cube所含的資訊量，因此以此為重點作為描述

---
#### 2) 讀者決定文章 (Longman)
> 撰寫一段文字，題目自訂，可以與自身研究相關。
> 設計：面向大眾/面向專業領域 之兩種文章

**題目：**

**科普取向：**

**Seminar:**

---
#### 3) 摘要拆解手術 (化)
> 尋找兩篇TOP論文，拆解其摘要。解構其目的、論數據、方法、結果等

**論文Ａ：** Hierarchical feature selection with multi-granularity clustering structure
:::info
Hierarchical feature selection addresses the issues caused by the presence of high-dimensional features in multi-category classification systems with hierarchical structures. Granular calculations are made to analyze the hierarchical relationships among categories when selecting the optimal feature subset. However, semantic hierarchy-based feature selection methods are prone to the semantic gap problem, which affects classification accuracy. In this paper, we propose a hierarchical feature selection method with a multi-granularity clustering structure that can effectively alleviate the semantic gap problem. Firstly, a hierarchical structure is constructed via bottom-up multi-granularity clustering based on feature similarities rather than semantic categories. This clustering hierarchy is conducive to solving semantic gap problems in the existing hierarchy. Secondly, the optimal feature subset is selected using the L1,L2-norms in each hierarchy’s granularity layer. This joint minimization approach can retain both the granularity layers’ shared features and granularity-specific features. Finally, we execute hierarchical classification according to the granular structure in a coarse to fine sequence. Extensive experiments demonstrate that the proposed method outperforms several state-of-the-art hierarchical feature selection approaches.
:::
#### Main Purpose
- What issue?
- Hierarchical feature selection addresses the issues caused by the presence of high-dimensional features in multi-category classification systems with hierarchical structures.
- Current solution of the issue?
- Granular calculations are made to analyze the hierarchical relationships among categories when selecting the optimal feature subset.
- Encountered Problems with Current Solution?
- However, semantic hierarchy-based feature selection methods are prone to the semantic gap problem, which affects classification accuracy.
- Proposed Solution (Main idea or novelty of this paper.)
- In this paper, we propose a hierarchical feature selection method with a multi-granularity clustering structure that can effectively alleviate the semantic gap problem.
#### Methods
- Steps to solve the problem?
- Firstly, a hierarchical structure is constructed via bottom-up multi-granularity clustering based on feature similarities rather than semantic categories. This clustering hierarchy is conducive to solving semantic gap problems in the existing hierarchy. Secondly, the optimal feature subset is selected using the L1,L2-norms in each hierarchy’s granularity layer. This joint minimization approach can retain both the granularity layers’ shared features and granularity-specific features. Finally, we execute hierarchical classification according to the granular structure in a coarse to fine sequence.
#### Results & Conclusion
- Brief conclusion
- Extensive experiments demonstrate that the proposed method outperforms several state-of-the-art hierarchical feature selection approaches.
> 這篇沒有提到result...為甚麼?
> - 因為這篇的result無法簡短地敘述完，所以直接下結論

**論文Ｂ：**
解構分段、邏輯
:::info
International environmental agreements have led to the need to reduce methane emission by dairy cows. Reduction could be achieved through selective breeding. The aim of this study was to quantify the genetic variation of methane emission by Dutch Holstein Friesian cows measured using infrared sensors installed in automatic milking systems (AMS). Measurements of CH4 and CO 2 on 1,508 Dutch Holstein Friesian cows located on 11 commercial dairy farms were available. Phenotypes per AMS visit were the mean of CH4, mean of CO 2, mean of CH 4 divided by mean of CO 2, and their log10-transformations. The repeatabilities of the log10-transformated methane phenotypes were 0.27 for CH4, 0.31 for CO 2, and 0.14 for the ratio. The log 10- transformated heritabilities of these phenotypes were 0.11 for CH4, 0.12 for CO2, and 0.03 for the ratio. These results indicate that measurements taken using infrared sensors in AMS are repeatable and heritable and, thus, could be used for selection for lower CH4 emission. Furthermore, it is important to account for farm, AMS, day of measurement, time of day, and lactation stage when estimating genetic parameters for methane phenotypes. Selection based on log 10-transformated CH4 instead of the ratio would be expected to give a greater reduction of CH4 emission by dairy cows.
:::
#### Main Purpose
- What issue?
- International environmental agreements have led to the need to reduce methane emission by dairy cows. Reduction could be achieved through selective breeding.
- Goal? (Main idea or novelty of this paper.)
- The aim of this study was to quantify the genetic variation of methane emission by Dutch Holstein Friesian cows measured using infrared sensors installed in automatic milking systems (AMS).
#### Methods
- Steps of producing this research?
- Measurements of CH4 and CO 2 on 1,508 Dutch Holstein Friesian cows located on 11 commercial dairy farms were available. Phenotypes per AMS visit were the mean of CH4, mean of CO 2, mean of CH 4 divided by mean of CO 2, and their log10-transformations.
#### Results & Conclusion
- Result
- The repeatabilities of the log10-transformated methane phenotypes were 0.27 for CH4, 0.31 for CO 2, and 0.14 for the ratio. The log 10- transformated heritabilities of these phenotypes were 0.11 for CH4, 0.12 for CO2, and 0.03 for the ratio. These results indicate that measurements taken using infrared sensors in AMS are repeatable and heritable and, thus, could be used for selection for lower CH4 emission. Furthermore, it is important to account for farm, AMS, day of measurement, time of day, and lactation stage when estimating genetic parameters for methane phenotypes.
- Brief conclusion
- Selection based on log 10-transformated CH4 instead of the ratio would be expected to give a greater reduction of CH4 emission by dairy cows.

---
#### 4) 審稿時光機 (0ya)
> 尋找一篇去年以前寫的一段文章，重新修改，並寫下修改原因。

**原始文章：** (2018 ISMAB)
:::info
Yellow sticky paper image samples are collected by Tainan District Agricultural Research and Extension Station (Tainan DARES) and Taiwan Agricultural Research Institute (TARI). Collection by Tainan DARES is done every two weeks from, September 2017 to December 2017, and cellophane was attached on top of the yellow sticky papers for scanning. Each scanned yellow sticky paper image has a resolution of 10200x14039 pixels using an Epson Perfection v39 table-top scanner. On the other hand, the samples from TARI are fresh insect samples that are forcefully stuck to the sticky paper after hatching the insect eggs from growth chambers. The two samples sets are different since the samples collected by Tainan DARES are very dusty since the traps are placed near crops and the TARI images are ideally clean sticky traps.
:::

**修改文章：**
Yellow sticky paper image samples ++were++ collected by Tainan District Agricultural Research and Extension Station (Tainan DARES) and Taiwan Agricultural Research Institute (TARI). The samples from Tainan DARES ++were collected from greenhouses++ every two weeks, and the samples from TARI were fresh insect samples that were forcefully stuck to the sticky paper after hatching the insect eggs from growth chambers. The cellophane was attached on top of the yellow sticky paper for scanning, each scanned yellow sticky paper image has a resolution of 10200++×++14039 pixels using an Epson Perfection v39 table-top scanner. The two sample sets ++differ in terms of their cleanliness++. The samples collected by Tainan DARES are dusty due to the placement of the traps near crops, while the TARI images are cleaner ++as they used the sticky traps in the laboratory++.

**修改邏輯：**
1. 修改過去式。
2. 調整順序，說明兩方的黏蟲紙皆是以相同方式掃描成影像。
3. 移除無用時間：考慮到後面會說到的樣本數量應更具有參考性質。

**有什麼新發現：**
　　其實應該是因為那時候ISMAB準備太趕，所以寫得很匆忙，而且因為是當時正進行的研究，會有一種默認讀者什麼都知道、但其實只有我自己知道的狀況。現在重新看過，感覺按照原本的寫法，會有一些誤導。剛好這段在描述從兩個地方、不同方法蒐集到的黏蟲紙，那就該說到二者的比較。因此發現Tainan DARES有寫到蒐集的時間段，但TARI沒有；並且將黏蟲紙掃描成影像檔的方法是接在Tainan DARES之後，這樣顯得只有Tainan DARES的黏蟲紙是以此方式處理，但TARI的卻沒有說明。原先想既然Tainan DARES有寫到蒐集的時間，是否TARI也應補上，但考慮對於樣本蒐集的步驟，樣本的數量較具參考性質，並且現在已無從考證TARI蒐集的確切時間，倒不如刪去。