四等獎

登革熱病媒蚊幼蟲主要孳生於人工積水容器中，幼蟲生長發育主要受到溫度與食物的影響。本研究至戶外調查人工積水容器，發現塑膠類人工容器為主要孳生類型。於實驗室以 11 種人工容器培養白線斑蚊幼蟲，結果發現塑膠底盆的幼蟲發育速度較快，蚊蟲平均翅長較長。以塑膠組成物質雙酚 A 進行試驗， 發現高濃度 (>50 mg/L) 雙酚 A 會導致幼蟲死亡， 活動力降低； 中濃度(12.5~1.56 mg/L) 會促進幼蟲生長速率，縮短發育時間；低濃度 (<0.78 mg/L) 則不顯著。以濃 6.25mg/L 雙酚A 處理蚊幼蟲，Q-PCR 顯示四個齡期的幼蟲蛻皮激素基因 (Ecr) 分別表現量都有增加，其中四齡幼蟲增加 9.68 倍，蛋白質分析顯示 34~72 kDa 之間的片段濃度增加。在蚊幼蟲防治上，4.0 %蛋胺酸和 1.0% 硼酸皆可 100%抑制孑孓活性，結合低濃度蛋胺酸(0.13%)和硼酸(< 0.5% )可以提升 30% 抑制孑孓活性的功效。

本研究旨在探索大型語言模型如何應用於音樂生成故事。研究動機源自音樂作為文化中不可或缺的一部分，但若要以文字精準表達出音樂中的故事情緒尚屬困難，藉由本研究提出的方法可以使故事顧及到音樂的情緒起伏。隨著 AI 的發展，我 們開始看到它們在各領域的應用。這項研究的目的是製作出一個系統能以音樂作為輸入，輸出音樂內的故事，為達成目的，我們結合多個模型。研究使用 PyTorch等工具，並探討文句和音樂的共同表示方法，實現情感匹配。研究結果顯示，音樂和文句情感辨識模型表現不錯，也研發出一個完整的生成流程。目前已有直接生成音樂的模型，也有把音樂統整介紹的模型，卻沒有依據音樂中的情緒生成故事的模型。我們研究就是在解決這個問題，結合到 LLaMA2預訓練模型生成出具情緒浮動的故事，要注意的是 LLaMA2的輸出限制最多只能有 4096個token。我們將此產生過程稱為 MTSPL (Music To Story Procedure with LLaMA)。

蕗蕎為原鄉部落常見且重要的香辛料作物，具有降脂、抗動脈粥樣硬化、抗菌等活性， 但對於血管舒張作用則少有相關研究。因此本研究以乙醇對蕗蕎進行萃取，再進一步以色層 分析等方法進行純化，其分離到之化合物由核磁共振磁儀（NMR）光譜技術分析及質譜(MS) 之鑑定，鑑定得到一個主要大量的固醇類天然產物: Laxogenin。另外透過頂空氣相層析質譜儀(HS-GCMS)，從生鮮蕗蕎(Allium chinense G. Don)共鑑定出 13 種化合物，其中包含 8 種含硫化合物，並以 Disulfide dimethyl 含量為 71%，占揮發氣體之含量最高。 在抑制血管收縮素轉化酶 (ACEI)活性測試中，取蕗蕎的酒精萃取物、大蒜的酒精萃取物、Laxogenin、diallyl disulfide 在濃度為 0.0125 mg/mL 下，其抑制率分別為 3.9 %、8.3 %、4.2 %、4.6 %。並透過電腦模擬分子對接技術(MDM)，發現 Laxogenin、diallyl disulfide 與 ACE酵素具結合親和性，以上結果顯示，推測蕗蕎具有抑制 ACE 酵素及降血壓之功效。

環境酸化為地球面臨的危機之一。面對酸逆境時，產氨及排氨為脊椎動物保守的適應策略，而本研究致力於探討脊椎動物之一的斑馬魚 (Danio rario) 在面對酸逆境時適應的特殊機制。首先，斑馬魚在pH值4.0環境下酸處理6小時後，水體及血液中氨濃度顯著上升。透過qRT-PCR，負責產氨的麩醯胺酸酶 (GLS, gls) 基因在鰓、肝、肌肉、腎中皆被誘發，顯示他們皆貢獻於排氨的過程。比較不同器官發現，鰓最早在6小時便啟動 gls基因，可知鰓除了負責魚類的排氨，也是面對急性酸逆境時最初的產氨器官。在斑馬魚的鰓上，我們也發現了硬骨魚特有的富含GLS的細胞 (GLS細胞)，再次應證了硬骨魚面對酸逆境時，鰓上獨特的適應機制。本研究為氣候變遷議題提供了非常有價值及參考性的資訊，也協助預測未來持續環境酸化下硬骨魚類的未來。

本研究探討之四氫吡咯酮作為一種常用的藥物骨架，其已被發現在多種藥理學方面具有應用價值。本研究所探討的反應結合了文獻中的含氮五元環合成方法以及另一篇文獻當中以雙氫鍵予體催化劑進行不對稱催化的原理，並利用計算化學的密度泛函理論對此反應進行熱力學分析。 本研究共分成四個階段，第一個階段為方法學測試，在此階段我們選定了以 ω-B97X-D 與 B3LYP 兩種方法進行後續實驗；第二階段則分析了本研究探討的四氫吡咯酮的各種異構物，並選定後續進行熱力學模擬的異構物；第三階段分析了此反應在不同溶劑下的可行性， 發現使用低介電常數溶劑(如：甲苯)對於反應可行性有正向幫助；最後一個階段則為催化劑改良，並發現到在催化劑的氫鍵予體旁加上拉電子官能基可降低反應活化能。在未來，我們將更進一步改良催化劑，試圖改變催化劑對各式立體異構物的反應可行性。

GROSSMANN, João Miguel Sastre. ReCiPla - Cyclic Soil Microplastic Remover: A way to remove microplastics from soil using electrostatics. 2023. 28 p. Research report – Scientific Apprentice Program, Colégio Dante Alighieri, São Paulo, 2023. Microplastics are the largest form of physical pollution on the planet. Affecting everything from terrestrial and aquatic environments to the air, compounds up to 1 micrometer in size are present inside the human body and can intoxicate the main organs in which they are found, such as the lungs, spleen, liver, and heart. Therefore, methods of removing these compounds from nature are essential, which is why this research is based on electrostatically removing MP from the soil. To this end, a vibrating conveyor belt was designed that would act in conjunction with a plate electrified by a Van de Graaff generator to separate the plastic compound using electric field induction. After characterization tests to quantify the voltage produced by the generator, which produced an average of 95 kV, the vibrating belt was made and will be used later in conjunction with the electrostatic method. This methodology suggests that it’s a success even after the electrified plate was applied to its structure. It carried out the proposed processes, such as moving the test masses, vibrating them, and fully supporting the electrified plate. In addition, the electrostatic removal method was tested to verify its efficiency and applicability. It was found that the removal of microplastics ranged it from 10 to 20% efficiency, suggesting it to be an effective method for separating microplastics. It should be noted that these statistics will be improved as the research progresses. In this way, the research proved capable of establishing an electrostatic removal method, as well as a process for transporting the material to be removed, thus achieving the objectives it set out to achieve. Finally, it should be noted that this research is still under development, with a view to applying the process in conjunction with the conveyor belt to carry out sample tests, as well as improving the removal process in the future to make it more efficient.

The purpose of the project was to find out whether humans can tell the difference between human-generated text and artificial intelligence (AI) chatbots-generated text and to identify how Al-generated text differs from human-generated text. The dependent variable was the results given by the participants (whether the paragraphs were Al-generated texts or human-generated text). The Independent variable was the participants in the experiment and the controlled variable was the type of paragraphs (both the Al-generated texts and human-generated texts) and time used to test each participant. The hypothesis for this experiment was that the participants were not going to be able to differentiate between AI-generated text and human-generated text. In this descriptive and mixed-method study, participants were presented with questionnaires. Each participant needed to state whether they thought each paragraph was human-generated or AI-generated. At the end of the questionnaire, the participants were asked to briefly explain what assisted them in differentiating between the two. They were given 60 seconds to decide. A stopwatch was used to time them. A sample of 456 participants took part in this project. They were not told how many AI-generated passages and human-generated passages there were in the selection. They only knew the total number of passages. Over 99% of the participants could not correctly differentiate between AI-generated text and human-generated text in all passages. Only four participants were able to get 100% of the questions correct. All four learners attend schools located in urban areas. After I did my analysis, I discovered that my hypothesis was incorrect. Four of the participants were able to get 100% of the questions correct this indicates that not all the participants were not able to tell the difference between Al-generated text and human-generated text. This rejects my hypothesis. However, the chances of humans differentiating between the two are very low. All four learners attend schools located in urban areas; this indicates that it is easier for learners attending urban area schools to recognise AI-generated texts than learners attending schools located in rural areas.

從家用微波爐到3C產品，皆有微波電磁干擾 (Microwave Electromagnetic Interference, MWEMI)的防蔽設計需求。依據2020年科展競賽作品中的薄層微量雙金屬催化技術可製得 Graphene/Ag-doped異質結構，但該作品未曾研究此材料的防蔽 EMI 能力。本研究優化此技術，成功將銀奈米結構沉積在石墨烯的缺陷及晶界邊緣，相關技術與實驗參數（化學氣相沉積法的加熱溫度、時長和通氣量）已發表於2023年科展競賽作品。本研究進階發現 Graphene/Ag-doped異質結構具高透明度和屏蔽 MWEMI 的能力，單層膜可屏蔽60%之 MWEMI，效能優於文獻上記載的2~3層石墨烯。而依據2023年文獻說明單層奈米碳管 (MWCNTs)異質結構經氟化處理後，可大幅提升 MWEMI 屏蔽效能，因此我們也將 Graphene/Ag-doped材料經由四氟化碳電漿處理，氟化後的屏蔽效能又比單層石墨烯-銀高出15倍，可遮蔽高達99.9%的 MWEMI。

在真菌釀造過程中除了釀出需要的酒精、醋、醬油，味增，釀造後都會產生廢料：酒糟或酒粕改良性質後測試有作為木質黏合劑的潛力。我們選出 6 種菌種，透過破碎、離心、鹼裂解、酸中和與離心濃縮做出蛋白質膠，小量黏合能力測試結果選最佳的紅麴蛋白質與米麴蛋白質。參照 ASTM 標準壓縮負荷與拉力負荷法。 1.壓縮負荷法結果顯示紅麴蛋白膠抗剪強度為 156.1kgf，米麴蛋白膠抗剪強度為 51.3kgf，為對照組 82.2%與 27.0%。 2.拉力負荷法結果顯示紅麴蛋白膠抗剪強度為 88.1kgf，米麴蛋白膠抗剪強度為 40.2kgf，為對照組 195.1%與 89.0%。 實驗結果顯示紅麴蛋白膠在壓縮負荷法強度接近對照組，拉力負荷法強度顯優於對照組，在作為木材黏合劑選擇上提供一種天然且無毒的選擇。

老化造成的神經退化性疾病，與細胞內粒線體機能異常有重要的關聯性。已知核醣核苷酸還原酶 M2B (Rrm2b) 的主要負責協助粒線體的基因合成、修復以及功能維持。本研究以核醣體核酸干擾技術，使 Rrm2b 表現量降低，探討其對不同種類中樞神經細胞粒線體 DNA 含量、細胞形態及功能的影響。結果顯示，Rrm2b 表現量降低並不會對粒線體 DNA 含量造成影響，但會使呼吸作用的效率降低，以及讓神經細胞更易分化，甚至反分化。同時，星狀膠細胞中的水通道蛋白表現量增加 165%，葡萄糖轉運蛋白表現量增加 27%，麩胺醯胺合成酶表現量則降低 22%。因此，Rrm2b 的低表現量，使星狀膠細胞運輸功能變強，而麩胺醯胺的合成變慢，則氨的代謝減緩。這些星狀膠細胞異常可能是造成腦部功能受損的原因之一。