Challenge Overview
Challenge Introduction
Challenge Goal
The goal of this challenge is to create ideas of Use cases of Domain-specific Semantic Search Function (DSSS API) built on the top of Fujitsu AI Platform, which is called ”Zinrai”.
We are looking forward to ideas which are creative, unique, competitive in the market and useful.
Background
About “Zinrai”
Fujitsu Human Centric AI, “Zinrai” offers a set of highly practical AI functions proven through more than 300 AI-related business projects and field trials, made available as a service in the form of APIs.
“Zinrai” are offered either in the cloud or on-premise, according to customer needs. So, customers can flexibly construct their system according to the confidentiality of the data to be handled.
What is the Domain-specific Semantic Search Function (DSSS API)?
Domain-specific Semantic Search API (DSSS API) provides abilities to retrieve document(s) from huge set of documents that matches with given keywords. A key differentiator from the simple keyword search is that DSSS API can retrieve documents based on how they are “similar or relative” to specified keyword.
The DSSS engine has its own mechanism to derive these “similar or relative” keywords from a given word. The DSSS engine manages words as vector information of multiple dimensions, and the similarity of words is expressed by the distance in the vector space. The DSSS engine automatically creates “word similarity information” by Machine learning process. It consumes limited domain-specific documents as training data and then calculates “word similarity” by the vector operation based on “word similarity information”.
The DSSS engine is differentiated from other search engines by the following points:
1. The DSSS engine automatically generates “word similarity information” from training data. Basically, the synonym dictionary is maintenance free, this makes it always available to search with the high accuracy.
2. In the similarity search which is based on a manually built synonym dictionary, word similarity in which there is no entry in the synonym dictionary are not reflected in the search. In the DSSS engine, since the word similarity is calculated by vector calculation, similarity of all words can be reflected in search
In addition, the DSSS engine provides functions similar to general search engines (like Google), but there is one very different point.
3. Because the DSSS engine has “word similarity information” for specific domain and performs tuning for specific domain, high search accuracy can be realized.
Please see the existing example section to understand how the technology is used today, and to derive your ideas.
The search function provided by the DSSS API is shown below.
Task Overview
You should recommend your ideas of Use cases of DSSS API. Ideas should be creative, unique, competitive in the market and useful.
And they are expected:
- to be a solution to issues in a particular business background (i.e. some business in some industry)
- to contain actors (i.e. users, operators, a service or application to utilize the API) and scenarios with them.
- to show how does the solution help to solve issues with utilizing the API in effective.
We also welcome suggestions for changes to the DSSS API.
Output
We are looking for ideas! - but we also need a structure to your ideas so we can understand them and make it easier when reviewing.
1. Title: Title of your idea
2. Description: High level overview/statement of your idea
3. Details:
3.1. Description of background and problem that you’re idea is going to solve
3.2. Description of how your ideas work to solve this problem
3.3. Advantages of your ideas
4. Key scenarios:
4.1. Actors (i.e. users, operators, a service or application to utilize the API)
4.2. Key scenarios for each use cases (i.e. How an actor uses functions of your idea)
5. Appendix (optional):
Format
1. A document up to 5 pages (2-4 pages are recommended) in word or PDF format to describe your ideas.
2. It should be written in English.
3. To use charts, diagrams and tables to explain your ideas is encouraged from from a comprehensive perspective.
Example industry areas
Examples of the industry areas the client is focusing on. (However, we welcome use cases for other industries.)
1. Health Care
2. Life Science (Pharmaceutical)
3. Retail
4. Financial
5. Manufacturing
6. Logistics
Examples of existing use cases of API
Several solutions have already been existing, which uses the API. Please note that the same idea as following will not pass the screening.
Example1:
- User: R&D division in a manufacturing company
- Document: Patent documents
- Problem: Keyword search can not accurately search patent documents. Search accuracy depends on the experience of the person in charge.
- Solution: By searching information that is semantically similar by API, researchers can easily find necessary patent documents
The figure below shows the search screen realized by the DSSS API.
“Concept word” means the main word constituting the document, and several words are selected automatically from the document by the DSSS Engine. “Search term” is the word that the user entered for searching.
Example2
- User: Automobile manufacturer
- Document: Accident investigation reports
- Solution: DSSS API helps users to do some investigations with the past accidents for specific purposes. It's a common situation where the same things are written in various words in many reports. With DSSS API, Users can retrieve related reports of one report they concern easily without preparing rich domain-specific dictionary.
Checkpoint
During this challenge, you can submit a checkpoint submission by Mar 20, 9AM EDT. Submissions after this deadline will not be considered as checkpoint submissions. This is optional, meaning even if you don’t submit to checkpoint you can still submit to the final round.
We are awarding 5 checkpoint prizes at $100 each, every member can win only one checkpoint prize and you have to submit to the final round to qualify for the checkpoint prize.
Judging Criteria
You will be judged on the quality of your ideas, the quality of your description of the ideas, and how much benefit it can provide for the client. The winner will be chosen by the most logical and convincing reasoning as to how and why the idea presented will meet the objective.
Note that, this contest will be judged subjectively by the client and Topcoder. However, the judging criteria will largely be the basis for the judgement.
1. Usefulness of ideas (30%)
1.1. How much benefit it can provide for users.
1.2. How much strongly needed in the personal lives? (It would be better if you can provide something that is “must need it” rather than “nice to have it”)
1.3. How much benefit your idea can provide in the corporate business? (i.e. It emerges 10 times productivity, It makes what was impossible ever possible, etc.)
2. Market advantage of ideas (30%)
2.1. How well differentiated from competitive? (Other search services provided competitors like IBM, Microsoft, etc.)
2.2. What is the appealing point to the customer?
3. Novelty and creativity of ideas (30%)
3.1. Uniqueness, Interestingness
4. Feasibility of ideas (10%)
4.1. Is it feasible? (i.e. It does not requires large-scale hardware or unavailable data such as state secrets.)
4.2. Is NOT it supposing something impossible to realize?Final Submission Guidelines
Submission Guideline
You can submit multiple solutions. But note that the same member can’t win two prizes at the same time.
Reference
http://www.fujitsu.com/global/about/resources/news/press-releases/2016/1129-01.html