How to use DAO to build a knowledge database with collective wisdom and sustainable incentive mechanism

Original author :   Eric Zhang

translator ：[email protected]

Reviewer ：DAOctor @DAOrayaki.org

original text :   Web3 Collaborative Intelligence – Knowledge Trees, Knowledge Forest, and Community Contributions

The original title ：《Web3 Collaborative intelligence —— Knowledge tree 、 Knowledge forests and community contributions 》

Special thanks Zeo、DAOctor、Zhengyu、Christina The contribution of 、 Review and feedback .

Constructing knowledge structure database and better visualizing knowledge are the key to advancing computer science 、 AI and Web The important task of . Before the world of cryptocurrencies and decentralized applications , old Web 3.0 The research mainly focuses on the construction of knowledge base and knowledge map , And representations based on these structures / Reasoning （ semantics Web）.

There are two general ways to build a knowledge base . One way is from Web And other data sources , Then organize them into the required knowledge database （ Mainly “ A triple ” or “ chart ” A huge collection of , And then execute “ Higher order logic ” Or machine learning reasoning structures and other intelligent tasks ）. Another method is to rely on human intelligence to build a database （ for example , We will discuss in more detail later Wikipedia、ConceptNet or Citizen Science project ）.

This article will first review some related innovations in the past few decades , Then discuss how we can move forward , Build a high-level knowledge database with collective wisdom and sustainable incentive mechanism .

The knowledge base 、 Knowledge map and Wikipedia

For a long time , People are interested in creating knowledge maps , There are two main reasons ：

1. The point connecting all the information and knowledge created by human beings ,
2. And implement reasoning and machine learning techniques on the knowledge map to produce better artificial intelligence , And use the system to improve Web2 The user experience of the product .

Now? , Obviously, most of the useful knowledge maps are used as Web2 Created by the basic tools of large and medium-sized companies . for example ,Facebook The knowledge map helps to better search Social Networks , The Google knowledge map helps to present relevant information . Because everything is closed , We don't know how the knowledge map is constructed , But from UI Look at , These knowledge maps will certainly help improve the user experience .

The efforts of the Wikipedia community are amazing . This is one of the first attempts to demonstrate the power of the Internet community . On the other hand , Open database can be used as Internet public goods . One example is DBpedia, It's one for wanting to take advantage of Wikipedia The application of the knowledge base provides API The database of . Another example is ConceptNet, This is a free semantic web , Can help AI and NLP Programs acquire general semantics .

However , How much can these Internet public welfare organizations do , There are some fundamental limitations . Wikipedia relies on donations every year , It's in one. 501(c)3 Operating within the organization , It's hard to put more advanced incentives on it and build a cooler infrastructure based on Knowledge Networks .DBpedia and ConceptNet And so on . As a non-profit organization , It is difficult for these public welfare organizations to build a community that constantly builds infrastructure and eventually forms an ecosystem . I used it in College DBpedia Of API Built a Wikipedia Graphical visualization and search tools . However , It was much harder to join a vibrant community . Now in the crypto community , The situation is very different , Developers with good ideas can participate in more activities , Team up and get the support of Multi Chain ecosystem .

however , I don't recommend building another Wikipedia（ also called DAO-ify Wikipedia, or “Web3 Wikipedia”）, Because despite the limitations of the current non-profit organization model , but Wikipedia The content and structure of the website have been well planned and organized , People have benefited greatly from its achievements . in general ,Wikipedia Good at storing knowledge description , And through Web1 and Web2 infrastructure , We have made knowledge searchable . What Wikipedia and the existing network infrastructure are not good at is presentation “ Human understanding ” Knowledge —— Structural knowledge in the human brain . To present this information , Human curation and human collaboration are the core , This is in Web1/Web2 Not well supported in the infrastructure , But by Web3 Infrastructure and coordination mechanisms will be available

** It is worth noting that , People try to build massive structured databases to enhance the understanding of knowledge . for example , image Cyc Such companies have been trying for decades to build a common sense knowledge base to help machines imitate the human brain . These companies eventually turned themselves into commercial software companies , Because powerful artificial intelligence obviously needs more than the knowledge base of nodes and relationships . Compared with building a structural knowledge base for machines , Human understanding of knowledge and human management are very important here —— Build a knowledge base of human understanding to help more people understand .

On the other hand , It's worth thinking about , How to add higher-level semantics to the current Web of Knowledge in , That is, the structural knowledge we describe in this article .

Citizen science and voluntary computing

Another branch of exploration I would like to mention is citizen science and voluntary computing . stay 2010 In the early s , There are many exciting projects in the scientific community , They use the wisdom of people to accelerate the progress of research and scientific discovery . There are generally two types of such efforts . The first is called voluntary computing , It assigns computing tasks to a group of personal computing devices （ for example [email protected]、[email protected]）. The second type is called citizen science , It creates repetitive tasks that everyone can perform （ This is not a derogatory term ！）. The project collects data from many contributors （ Sometimes it is the result of analysis ）, And input them into some research projects to create meaningful results （ for example , stay Citizen Cyberlab、SciStarter Or projects listed in the machine learning community , Tagging images to enrich training data can be crowdsourcing ）. Without inventing “DAO” Think of these efforts as “DAO”, The coordination of decentralized communities is nothing new ！

Many projects have been successful , But unfortunately , The sustainability of these projects is again Limited [email protected] No longer in operation , Many citizen science programs could have lasted longer, but they didn't . Incentives and ecosystems are two important aspects of any collaborative effort . There is no ecosystem , Innovation will be limited . There are no sustainable incentives , There is no vibrant community , There will never be an ecosystem .

The structure of complex concepts and knowledge

Now let's consider what high-level concepts and knowledge are like . Intuitively , When we “ understand ” When a concept , In fact, we understand quite a few details of this concept . We can think in two ways “ understand ” The process of ：

1. Understand through tree structure

When we try to “ understand ” something , Or say “ Study ” When doing sth , We will decompose it into a tree structure . for example , If we want to understand “Merkle Trees ” This concept , We must understand “ Password hash function ” and “ Tree data structure ” Such a sub concept , This requires us to further understand “ hash function ”、“ Resistance to PengZhuangXing ” And so on .

The deeper the tree breaks down , The more primitive the concept is . At some point ,Web There will be some very direct resources that can be referenced directly （ for example , Wikipedia pages or articles / video ）.

The concept of “ decompose ” It's a tree structure

We can start from the old AI Find some similar ideas in .K Line theory shows that our memory and knowledge are stored in a tree structure （P Nodes and K node ）. Although there is little evidence that this structure does exist in our brains , But the model has the ability to explain human memory and how the human brain works , The tree structure is indeed the most concise form of storing structural knowledge .

We can use a tree structure to store and understand both directions —— Decompose and build .

If we want to retrieve details , We decompose a knowledge tree . On the other hand , If we have a knowledge tree , We can use this tree to build larger trees （ That is, the higher abstraction of knowledge and understanding ）.

Use concept _2“ establish ” Concept _1

stay “ structure ” Under the circumstances , have access to “Merkle Trees ” Trees are used as nodes to build more complex knowledge trees , for example “Verkle Trees ” or “Merkle Multiple proofs ”.

It is worth noting that , The key point here is the structure of the tree . Knowledge tree from root concept to leaf , Point to all pairs of existing Web Necessary references to resources . The relationship between nodes is not important here （ And knowledge map system “ triple ” Different thoughts ）.

2. adopt “ Related knowledge ” understand

We also add more “ Context ” To gain a deeper understanding of knowledge . just as Weigenstain The famous saying of ,“ but ‘ 5、 ... and ’ What does the word mean ？ There is no such problem , Only “ 5、 ... and ” How is the word used ”. The idea behind it is , The meaning of something actually depends on other concepts related to it , Together they determine the meaning of something . By adding more context （ That is, the relevant knowledge of knowledge itself ）, We can be more “ thorough ” Understand knowledge well .

Generally speaking , People understand trees more easily , Not a picture . Instead of building a knowledge map , How about putting “ Related knowledge ” Think of a more practical way —— A set of knowledge trees connected by root nodes , In essence, it forms a knowledge forest .

Knowledge forest can be built as a database of many knowledge trees （ Parallel planting ）. We can perform two basic operations on the database .

1. Make connections between different trees . When we visualize the knowledge tree , It will be very useful .
2. The feature of knowledge tree can be constructed as a vector in a vector space . You can then use vectors to correlate conceptually related but not through (1) Directly linked knowledge tree .

Measure the relationship between knowledge trees

About the depth of understanding

Generally speaking , People have different levels of understanding of the same concept . For some people ,Merkle The concept of a tree is simple , No further decomposition is required （ Their brains have encapsulated this concept into some common sense ）, Others don't have enough information to understand “Merkle Trees ” The concept of , A further breakdown may be required .

therefore , Knowledge trees need not be mutually exclusive , This means that there may be overlap between different trees . There may be trees that explain basic concepts , And trees built for high-level concepts .

Overlap may create redundancy between trees . To reduce redundancy , We can introduce the following operations ：

1. Cross tree references （ Dashed Links ） - Create a link , Connect nodes from one tree to the root of another tree .
2. Merge - There may already be subtrees under the nodes of the two trees , If the basic tree has not covered some valuable nodes 、 Leaves and references , It might be worth merging information from a higher-level tree into a more basic tree .

Cross tree reference links

Merge two trees into one

Knowledge trees and meta operations

A single knowledge tree consists of a root 、 A set of child nodes and a set of leaves , Organize into a tree structure . Then we can define a set of basic operations to create and refine a tree .

1. Create a root （ Trees ）
2. Add child nodes
3. Add leaves to nodes
4. Add reference links to leaves

Then we can define a series of advanced operations for actual users to “ planting ” One tree and contribute to one tree .

1. Add subtree - Introduce necessary child nodes for the knowledge tree with complete nodes and leaves
2. Merge two trees with the same concept

Knowledge forest

Plant a large number of knowledge trees , We have a forest of knowledge ！

Knowledge forest is a large group of knowledge trees planted together . An interesting fact about the knowledge forest is , There may be entanglement between trees . Theoretically , The connection between different nodes and leaves can be arbitrary （ for example , A link between the leaves of one tree and the roots of another tree ）. actually , If we add dashed Links , Knowledge forest “ somewhat ” It becomes a knowledge map . however , What matters is the personal knowledge tree .

for example , The dotted line shows MACI Trees and zk-Snark Links between trees .

The leaves of the knowledge tree are connected to the existing articles on the network / video / resources . therefore , The layer above these leaves is the structural information or understanding layer .

What we can do with the knowledge forest is completely open . Perhaps the most important thing we should consider is the ecosystem of collaborative knowledge bases from the beginning . We may want to do a lot about the knowledge forest , Here are three examples ：

1. Visual knowledge tree and knowledge forest
2. Browse the knowledge forest through dashed links
3. Find the knowledge tree cluster

Build a DAO, Not a non-profit organization

Nonprofit organizations can make things happen , but DAO Can make things better . The idea here is to map a set of tree operations to a set of incentives . The more standardized meta operations are ,DAO The more extensible the coordination members are .

Knowledge tree operation <-> DAO contribution

In the case of knowledge trees ,DAO Contributors to can create a root （ be equal to “ establish / Plant a tree ”）, Add a knowledge path （“ Planting trees ”）, And add reference links to the leaves . The incentive mechanism creates a set of rules to reward community contributors who take verifiable actions to plan and plant knowledge trees .

meanwhile , The Review Committee （ Or review groups ） It is also important for planning and quality control .DAO Coordination and motivation have been extensively tested （ for example ,DAOrayaki DAO）, And a similar structure can be implemented here .

Knowledge forest and knowledge map

When we learn new concepts and acquire knowledge , Trees are easier to understand . For any particular subject , Human beings can easily understand the knowledge structure in the tree , Because there is no loop in the tree , If the depth of the tree is limited to a certain level , It's much easier for the brain to process and remember .

Besides , The representation of knowledge graph is limited in representing fuzzy or fuzzy connections between knowledge nodes （ Common sense knowledge represents the same problem ）.

This does not mean that knowledge trees are always better than knowledge graphs . In storytelling , Knowledge graph is better than knowledge tree （ for example , Pictures of all Greek myths ） More useful . There are actually many existing tools （1 2） To build the knowledge map , But to my surprise , Most of them are becoming SaaS company .

A dedicated to the practical implementation of knowledge trees and knowledge forests BUIDLers The team has many details —— data structure 、 The product design 、 Details of contributions and incentives 、UI wait . For all that , If we want to build a knowledge forest , In general, I think it should be a public product to organize knowledge and open to all people in the world . however , Let's see Dora What the community came up with ！

Conclusion

This idea is based on the existing Web infrastructure （ Like Wikipedia, etc ） Build a new knowledge base , And make it available to all , So as to minimize the complexity of understanding abstract knowledge （ By like Web Or a knowledge map like Wikipedia can be as complex as O(nlog(n)), But there are n A tree of nodes has only log(n) The depth of the , This makes navigation easier ）. And DAO Contributor coordination in , And use advanced encryption native incentives to ensure the sustainability of the organization . The ideas in this article are not complete , There is still much room for discussion and improvement , If a team wants to make it a reality , There are many engineering and product issues to consider .

reference

1. semantics Web：https://en.wikipedia.org/wiki/Semantic_Web
2. A triple ：https://conceptnet.io/
3. Higher order logic ：https://en.wikipedia.org/wiki/Cyc
4. ConceptNet：https://conceptnet.io/
5. DBpedia：https://www.dbpedia.org/
6. Wikipedia Graphical visualization and search tools ：
7. https://github.com/zhangjiannan/Graphpedia
8. Multiple ecosystems ：https://hackerlink.io/grant/dora-factory/top
9. Cyc：https://en.wikipedia.org/wiki/Cyc
10. Volunteer Computing ：https://en.wikipedia.org/wiki/Volunteer_computing
11. [email protected]：https://lhcathome.cern.ch/lhcathome/
12. [email protected]：https://setiathome.berkeley.edu/
13. Citizen Cyberlab：https://www.citizencyberlab.org/projects/
14. SciStarter：https://scistarter.org/
15. Knowledge map building tool 1：https://obsidian.md/
16. Knowledge map building tool 2：https://www.ideaflow.io/

DAOrayaki DAO Research award pool ：

Funding address :  DAOrayaki.eth

Voting progress ：DAO Committee 2/0  through too

Total reward ：130USDC

Research types ：Web3, Knowledge Trees, Knowledge Forest,Community Contributions