Bibsonomy
Explore and reserach 200M+ papers and publications with natural language understanding and manage them with Bibsonomy.
The Bibsonomy ChatGPT plugin helps you research scientific publications, manage your Bibsonomy library, and analyze its content. You can search for, connect, add, and edit publications on Bibsonomy with specific filters. It also supports rudimentary bookmarks and emphasizes the importance of using relevant tags. To ensure accuracy, user data and tags should be obtained before adding a publication. It's important to note that adding and editing publications directly impacts the Bibsonomy service, so data should be presented for user approval. Use this plugin to access scientific publication data and combine it with user input or other plugins designed for scientific publication search.
How to
Comments (0)
API docs
Learn how to use Bibsonomy effectively! Here are a few example prompts, tips, and the documentation of available commands.
Example prompts
-
Prompt 1: "Search for peer-reviewed articles about climate change."
-
Prompt 2: "I have found an interesting paper on climate change, how can I add it to my Bibsonomy library?"
-
Prompt 3: "Can you help me edit the tags of a publication in my Bibsonomy library?"
-
Prompt 4: "Find the most recent publications related to quantum computing."
-
Prompt 5: "I found a paper on AI in healthcare, how can I add it to my Bibsonomy library with the correct tags?"
Features and commands
| Feature/Command | Description |
|---|---|
searchPublications | This command allows you to search for relevant publications based on a set of keywords. You can specify the sort order, filters (such as peer-reviewed articles), and the range of publication years. |
addPublication | This command adds a publication to your Bibsonomy library. It is important to provide accurate and structured data for approval before executing the command. This includes the title, authors, abstract, tags, and other relevant information. |
editPublicationTags | This command allows you to edit the tags of a publication in your Bibsonomy library. You need to specify the publication's unique identifier (hash) and the updated tags. |
searchSemanticScholar | This command performs a search using the Semantic Scholar search endpoint. It can be used in combination with the Bibsonomy endpoint or on its own to retrieve more comprehensive search results. |
For AI
Namebibsonomy
DescriptionPlugin for scientific research and publication management that connects ChatGPT and the Bibsonomy bookmarking and publication management system. It allows users to search for scientific papers in the Bibsonomy and Semantic Scholar databases, add them to their personal Bibsonomy account as posts, edit existing posts, and retrieve and make connections between posts to build a knowledge graph. When searching for publications, use your advanced natural language processing skills to mirror a human researcher's approach to investigating a new topic. Use your broad knowledge of the world to understand the user's research intent and generate appropriate search queries. These search terms can include synonyms, related concepts, common abbreviations, and their full forms to allow for catching as much relevant results as possible. While this broadens the search it should stay within the specific requested topic, since the emphasis is on providing multiple specific search terms to ensure relevance of results rather than over-generalization. To avoid over-generalization use different method calls for different topics. The goal is to capture the essence of the user's research intent to ensure a comprehensive and relevant search experience. When delivering search results to the user, it is critical to use only the information provided by the search endpoints and not add your knowledge to it, as absolute accuracy is key in scientific research. You are however welcome to and encouraged to narrow down the search results further and/or re-rank them based on your understanding of the users' research intent to only deliver the most relevant results in their optimal form. It is considered best practice to store relevant research results in the user's Bibsonomy account so that these publications can be referenced later. Since adding publications to an account causes changes to public databases, these additions can only be made with the explicit permission of the user. When adding entries, the relevant paper metadata can usually be retrieved automatically by simply specifying the paper ID (main_id) and the service from which the paper was originally retrieved. Since these IDs must match the databases exactly, it is crucial to use the exact IDs provided by the plugin responses (or the user). These IDs must match exactly and are case-sensitive! Because of that, ensure that they are taken exactly from the users' input or (more likely) from previous plugin responses. In case of a 404 error, try again to get the exactly correct ID from the chat context. Additionally, posts must be created with appropriate tags. Since tags are a powerful management system within a user's account, they should keep a similar style and be reused when appropriate. To enable this, before adding the first post to Bibsonomy, ALWAYS fetch the tags the user used before by calling the getBibsonomyUserInfo endpoint! Finally, a brief description of the post must be provided. To gather useful information for this description, the getDataDetails endpoint can be used, which returns an abstract and TLDR of the publication, if available. This description can be biased towards the users' research since it explicitly differs from the abstract of the paper and can be personalized. It should not just be a trivial statement obvious from the papers title and should not recite the bibliographical information since this is stored separately! However, it should still be related to the content of the Paper/Bookmark. Given Bibsonomy's role as a scientific database, it is crucial that all data be accurate and not artificially generated by language learning models such as ChatGPT. If data is not available or appears to be incorrect, always ask the user instead of creating imaginary data. If it's unclear, if the user wants to create a Post or a BibTeX (Publication) rather ask then create a false entry type, since changing them afterward is not possible. Usually, a BibTeX entry is the more probable intent, since Bookmarks can't really store any metadata. If there are problems with the plugin (like failed method calls or misalignment between the plugins functions and the user use-case) ask the user to provide feedback quickly or do it automatically. This is a research project and the best way to improve it.
Updates
First added1 August 2023