I prefer tools that remain ‘pure’. They should focus on their core functions, allowing users to complete tasks with a smooth learning curve, rather than expanding infinitely through plugins to achieve functions beyond their core purpose. For example, using Obsidian plugins to implement complex features that Markdown does not natively support. For me, Zotero’s core responsibility is to efficiently manage the metadata of my references and provide necessary assistance during reading.
Recently, I discovered that nearly a quarter of the entries in my library had incomplete or incorrect metadata. So, I took some time to tidy it up and used this opportunity to document my current Zotero configuration for future reference.
A foreword: We sometimes fall into the ’tool trap’. With an ever-increasing number of products, it seems every new app wants to be associated with AI to attract more investment. This inadvertently creates a kind of tool anxiety, as if we’ll be left behind if we don’t master a certain new tool. However, in most cases, this is not true. The majority of tools only offer minor assistance, and the time spent tinkering with them often outweighs the actual benefits they bring. Installing a hundred plugins for Zotero will not lead to a significant increase in efficiency.
That said, tinkering with tools is, for me, a form of positive reinforcement. Just like fiddling with screwdrivers and spanners, playing with these digital tools gives me a sense of accomplishment. Therefore, spending this weekend on this is more meaningful than lying in bed scrolling through short videos.
File and Data Syncing
Zotero’s synchronisation is split into two parts: data and files. For data syncing, you just need to register and log in to a Zotero account. As for file syncing, since Zotero only provides 300 MB of free storage, I opt to use the WebDAV protocol. I tried using a NAS, but considering the convenience of external network access, I eventually chose Jianguoyun. Jianguoyun offers free users 1 GB of upload and download traffic per month, and the speed limit is sufficient for daily use. Although you might hit the traffic limit during the initial sync of a large number of files, subsequent incremental syncs are generally problem-free.
Some PDF files can be unusually large. For instance, I once came across a paper of just over ten pages that was 48 MB. This is usually because the images within it have not been compressed, for example, high-resolution original images have been embedded directly. In such cases, you can use Adobe Acrobat’s file optimisation feature, setting the maximum image PPI to 250-300 and the JPEG compression quality to ‘High’, to significantly reduce the file size.
The Plugin App Store
The traditional way to install Zotero plugins is to go to a webpage, download an XPI file, and then manually install it through the settings menu. This plugin, however, provides an integrated app store that supports one-click download, installation, and updates. It makes managing plugins as simple as installing an app on your phone. All the plugins mentioned below can be quickly installed through it.
High-Quality Translation
This is perhaps the most-starred of all Zotero plugins. Its function is simple and direct: it provides translation for selected text within the Zotero reader interface. The plugin is actively maintained and integrates various translation service APIs, enabling high-quality and fast translation of academic papers.
LLMs are currently one of the best methods for translation. They can better understand context and provide more accurate translations. In particular, models with short CoT manage to balance both speed and translation quality. It’s worth mentioning a specific model within the LLM family called Qwen MT. It focuses on translation tasks, and its API differs from normal OpenAI completion or others. This plugin has been specially adapted for it.
PDF Downloader
This plugin allows you to say goodbye to the tedious process of copying the DOI, going to Sci-Hub to download the PDF, and then manually adding it to Zotero. After you add a reference entry, it will automatically try to find and download the corresponding full-text PDF.
A Linter for Metadata
Think of it as a ’linter’ for your Zotero metadata. It can retrieve, complete, and format all metadata fields using a DOI or ISBN, thus maintaining the integrity and tidiness of your library’s information.
Additionally, it offers a feature to fill in journal abbreviations to meet the requirements of specific reference formats. For Chinese content, it can also populate the location field based on university names to comply with citation standards like GB/T 7714-2015.
Jasminum
This plugin is mainly for handling Chinese literature. It can retrieve metadata from CNKI and generate a table of contents for Chinese PDFs. Even if you mostly read English papers, it is particularly useful for managing theses and dissertations from Chinese institutions, as it can accurately fetch information and create a table of contents.
High-Focus Reading
Bionic Reading is a method designed to help non-native speakers improve their reading efficiency. It works by bolding the first few letters of each word, guiding the visual focus to allow for more efficient reading of English text.
While similar extensions exist for Chrome and work well, this plugin’s application to PDFs is less than ideal, perhaps due to parsing issues. I only enable it occasionally when I’m struggling to concentrate. But usually, in such situations, putting the reading aside and doing something else might be a better choice.
A Not-So-Awesome GPT
This plugin is like an LLM chat tool with integrated preset prompts. It allows you to ask questions about the current paper within Zotero, such as asking for a summary. It uses RAG technology behind the scenes, but the PDF parsing, vector embedding, and question-answering performance are not ideal, and the user interface could be more intuitive. In comparison, you often get better results by feeding the PDF to more specialised products like NotebookLM. In my actual use, its core function has been reduced to almost exclusively ‘summarise the full text’. Its other features are also not as user-friendly as the text-selection features in Cherry Studio.
Returning to the Essence: Focusing on Reading
After all that, the fundamental reason for using Zotero in research and study is to better manage and read articles. No matter how advanced the management and retrieval tools are, or how good the auxiliary and translation methods, they are useless if you don’t read. Installing a hundred plugins in Zotero and collecting a thousand beautifully organised papers in it won’t increase your actual knowledge one bit. Therefore, once the tools are set up, let’s forget about them and return to reading itself.
As Gemini says: After all, tools are just the boat and oars; we still have to row ourselves to the shores of knowledge.