Слайд 2Experience
Financial Companies: «Diasoft», «BCS», «Sberbank Technologies»
Databases
Spring Open Source | Java
RedHat OpenShift
Jenkins Jobs
| Versioning | Checking
Models
Слайд 3Objective. Huawei approach
Science content oriented search improvement
+ Marked Data
+ Reducing
amount of labels
+ Data validity
+ Model Quality Improvement
- Label inconsistency, Poor data
Слайд 4Model chosen and its training problems
New tag, correctness and matching
Results:
Ordering and targeting
Tag weight definition, Mind Map
The likelihood and consistency of tag content (and more parameters)
Слайд 5Convenience is an idea
Machine learning and working at the same time
Connection
to new databases via a separate simple module
Scanning, preprocessing tags as an initialization process
Mixed search sources
Слайд 6Future goals
Analysis of difficulties and "dead ends" in research work, including
the hierarchy of problems
Multilingualism (only English is taken as a source for the current set)
Processing scientific formulas (Specific for particular area of Science)
Formation of general templates for an arbitrary field or Science