We provide consulting and software services to help you integrate and curate your content. Our services may include providing access to our unique technology, "the Networker".
· Data Integration, Mapping
· Knowledge Graph Curation
· Taxonomy/Ontology Management
· Metadata Refactoring
· Spreadsheet Replacement
· Writing tool
· Web site production
· Two user interfaces: Dialog boxes and local graph visualization.
· Collaboration Tool
· Current Usages
Data Integration and Mapping
The Networker can be used to import data from various sources, and integrate by mapping its components to high-level topics. Import is created via custom scripts that extract semantic information by various means, transform it according to specific rules when necessary, and import them into the Networker using the Networker’s API.
Knowledge Graph Curation
Every information item is stored as a topic. Every topic can have several names, can be related to other topics via semantic relationships, it can have links to relevant URLs, and can have a description attached (that can be as long as necessary).
Different topics can be merged if they were created independently but represent the same meaning. One topic can be split into two (or more) different topics to isolate specific meaning for each of them. The same name can be used to designate different topics with a disambiguation feature. Relations between topics can be modified, either by changing the value of their semantic or by changing the origin topic or the target topic.
The Networker can be used in a multi-lingual environment, using the fact that topics can have multiple names.
A relation type, and a disambiguating scope, are treated as plain topics. Therefore it’s possible to alter their names, add synonyms or other language equivalents.
Any instance of a relation is also treated as a topic. Relations can be documented, and can be nested, enabling for comments and annotations about the context and the provenance of the information.
The Networker contains several ways to access information. A simple search box helps finding topics based on their names, an alphabetic index is available to scan the existing list of topics (as well as a consolidated list of all topics present). Furthermore, relation instances are listed by selecting a relation type.
Taxonomy/Ontology Creation and Management.
Any relation type can be used to relate topics. A built-in relation type (broader / narrower, or parent / type, or category/instance) has been given a specific user interface to facilitate maintenance of taxonomies based on hierarchical classification of information. This feature can be enabled in the Settings. This feature enables to quickly create or manage hierarchical relationships used in many contexts. Lists of categories are available in the index pages.
The graph-based organization of data may lead to simplify the management of metadata, especially when integrating information from different sources. When rules apply on what type of data can be connected to another type, the fact that this connection exists implies that the origin has to be of a certain type, and the target as well. For example, if the relation describes the fact that a person works in a company, the “works in” relationship origin is always a person and the target is always a company (or an organization). The categorization of the origin topic as a person is therefore redundant and can be removed. The same holds for the categorization of the target topic as a company. In this context, a person is (among other things) somebody who works somewhere, and the company is a workplace. This may actually help further distinguish the workplaces into “companies”, “organizations” , “charities”, “governments” or whatever further characterization may be relevant in this context.
Spreadsheets/Database Tables Replacement
Spreadsheets were designed for numbers, but are also used for data in tables, as a light-weight solution for a relational database table. When spreadsheets are used to organize individual data, they work when every item can be unambiguously characterized as part of a column with a given header. There are several weaknesses to this approach: 1. Data is not always characterized exactly as an item belonging to the category defined in a header. 2. There can be many redundancies (for example, City / Zip code/ Country: once a zip code is entered, with the corresponding city and country, it should be the same for every other occurrence, but it’s repeated on each row where it appears). 3. Data doesn’t get connected at the cell level. If the same content (for example “Chicago” appears in different rows, but also possibly in different columns), it’s hard to view all information about Chicago because it is spread out in many different places.
The Networker has an interface that is as easy (or even easier) to use as a spreadsheet and has
been explicitly designed for ease of use. It is more flexible, and since everything is a topic (a header as well as a cell content), every single piece of information can be related to any other.
The Networker can be used as a writing tool. Considering that a text is structured as made from sections that can have any number of subsections that have paragraphs, it is possible to create topics using the taxonomy tool (narrower/broader) and create other subtopics at any level desired. Every topic has a field for free text (HTML) with an editor.
If the top level topic is named “Contents”, then using the “Export Content Tree” features will automatically produce a document containing the various sections and their contents into HTML, that can be easily transformed into another format, such as docx, markdown, etc., for further editing. Our experience so far is that this tool is useful in order to organize a first draft of a written text project, and later switch to a traditional writing tool, when material has started to accumulate.
Web Site Production
Among the export features, exporting to HTML has the effect of producing a web site that has one page per topic, and shows all the links between the different topics. It is possible to modify the stylesheets (CSS) to customize output according to every specific context. The Networker can therefore be used as a quick way to produce a web site with many interconnected pages. Generating a new set of static HTML can be done at any moment, and the process of posting to a given server can be automated.
Other export formats are available, such as JSON, XML, SQL, CSV, XLSX, and can be customized per usage.
Two User Interfaces
The Networker has been designed with two user interfaces that can be used together. Every user has the ability to select one or the other (or both) depending on what’s considered optimal for productivity. One contains dialog boxes with autocomplete text input. Each topic already present in the database is shown as a possible choice, and it is also possible to create new values at each input box. Every dialog box contains specific features for the component:
Topic names, relations, broader/narrower, bookmarks, notes, and deletion.
The other possible way to interact is through a graph-based view. Instead of showing the whole graph, which makes it almost impossible to interact, the choice is to view either the graph of the immediate connections with each topic (zoom level 1), two degrees of connections (zoom level 2), and three degrees of connection (zoom level 3). It is possible by clicking on any node to display the graph that is centered on this node, and therefore to navigate the graph when necessary. If the number of connections to a topic is higher than a predefined threshold (currently set at 20), then the connections are not shown in the graph, but are still showing up in the dialog box for relations. This interface has been designed to be usable in most cases.
The Networker is a cloud-based service. Every customer has their own dedicated server containing their data and have their own private access code (password protected, or possibly ssh-enabled passwordless access). The static HTML web site and other export formats can be posted on a password protected web site or on public web sites depending on the requirements.
It can be used by several people sharing access to the same data, and therefore is a collaboration tool.
· Research Paper and Fact database
· Family Tree
· Contact Database
· Thesaurus Management
· Software documentation
· New York Subway Map
Import / Curate / Export
Our leading edge technology, The Networker, is a Swiss army knife for Knowledge Management. It is designed to be respond to the following business needs:
- Integrate legacy and current data in multiple forms with custom scripts.
- Express information that doesn't fit into rigid schemas (databases, XML, spreadsheets).
- Curate inaccuracies in the results produced by automated processes.
- Enhance the productivity of Subject Matter Experts with an easy to use interface, with zero learning curve.
- Empower developers to access the full power of a programming language (Python) to access the API.
- Enable recurring content cleaning and therefore remove the need of obsolescence.
The Networker is a cloud-based service made of import/export custom scripts and a curation interface.
Integrating various sources into a knowledge graph is not only a question of reading files in different formats. It also means that customized integration models have to be defined. For example, a spreadsheet containing three columns: First Name, Last Name, and Company can be imported by creating one topic for each person (whose name is acquired by grouping their first name and last name) and creating a link whose semantics is "works for" to the data in the corresponding column for Company.
This is why every import needs to be automated with a script that reads the input data and populates the topic database with a script that uses the API.
The Networker container several techniques for curating data, which are presented in the demo available below:
- Creation of new topics
- Creation of relations to new topics with user-defined roles. Related topics can be created on the fly.
- Split a topic into different topics, assigning properties to each derived topics.
- Merge two topics into one.
- Ability to designate a topic with different names
- Ability to use the same name for different topics, via a disambiguation mechanism.
- Ability to change relation semantics, as well as the order in which topics are connected.
- More on essential features.
The Networker contains the ability to create a static HTML web site, which is a read-only version of the data contained in the database. The demo features a straightforward export of the data. However, filters can be set to only export a part of the data and can be done on a customized basis.
The export to JSON or XML can also be customized to contain data presented with object or element names that are directly usable in other applications.
Other export can be provided depending on the formats and data structures for specific environments.
We help you design the structure of your content by creating XML schemas.
Powerful XML applications can be created using XSL transformations.
We help you track the provenance of your content with our "Data Projection" approach. Use it to improve the efficiency of your collaborative editing environment, and the security of your information system.