No articles match
Statistical profiling of bibliographic metadata with marcxmlr2 days ago
A catalogue as a statistical population | Dataset and limits of inference | 1. Coverage: how often does a record have topical subject access? | 2. Indexing intensity: how many topical subject fields per record? | 3. Authority identifiers: use field instances, not subfield rows | 4. A more revealing association: subject subdivisions and $0 | Check an important confounding factor: subject vocabulary | 5. Return from a statistical pattern to original MARC fields | 6. From a sample to a catalogue-scale study | Conclusions
Architecture and validation17 days ago
Semantic reference first | Canonical contract | Parallel execution | Validation before package conversion | Package conversion rule
Getting started with unknown XML17 days ago
The core idea | Inspect a sample | Define the profile | Rectangle the XML | Parallel execution is an option, not another workflow | XSD-assisted review | Exploratory analyst table
Large XML, streaming and parallel execution17 days ago
Compile once | In-memory execution | Streaming execution | CSV output | Parquet output | Automatic workers and scheduling | The two parallel strategies | parallel_chunks | shared_chunk | Why streaming auto does not pre-count the XML | Correctness invariants
Architecture and validation1 months ago
Semantic reference first | Canonical contract | Parallel execution | Validation before package conversion | Package conversion rule
Getting started with unknown XML1 months ago
The core idea | Inspect a sample | Define the profile | Rectangle the XML | Parallel execution is an option, not another workflow | XSD-assisted review | Exploratory analyst table
Large XML, streaming and parallel execution1 months ago
Compile once | In-memory execution | Streaming execution | CSV output | Parquet output | Automatic workers and scheduling | The two parallel strategies | parallel_chunks | shared_chunk | Why streaming auto does not pre-count the XML | Correctness invariants
Getting started with summarisebig2 months ago
What problem does summarisebig solve? | 1. Let Arrow do everything when it can | 2. Go as far as possible in Arrow, then finalize in R | 3. Arbitrary R functions: parallel_chunks | 4. Lower-memory materialization: shared_chunk | Choosing a strategy