To translate an IVOA VOTable 1.5 quickly, leave its column identities, numerical values, data types, units, Unified Content Descriptors, coordinate references and serialization intact, and translate descriptive metadata only through a format-compatible and documented route. Human readers can see translated explanations of a star catalog, but a query tool must still interpret every quantity and coordinate as the same scientific measurement.
VOTable translation is not ordinary spreadsheet localization. Astronomical catalogs often contain columns that look like small strings but carry strict scientific meaning. A column named ra can describe right ascension; a field carrying unit="deg" must not be relabeled as hours without an actual numerical conversion. The translated interface needs to make the science easier to understand while preserving what the instrument, model or catalog actually recorded.
This guide uses the International Virtual Observatory Alliance’s VOTable 1.5 Recommendation dated 16 January 2025. It covers <VOTABLE>, <RESOURCE>, <TABLE>, <FIELD>, <PARAM>, <DESCRIPTION>, Unified Content Descriptors, units, coordinate and time systems, data serialization, missing values, catalog provenance and multilingual display. Sample XML is checked for well-formedness; full VOTable schema and application behavior require additional validation.
The central distinction: measurements are evidence, labels are interpretation
A table can be useful without being linguistically friendly
An observatory catalog may include columns for right ascension, declination, flux density, uncertainty and source classification. These values let astronomers perform numerical queries, cross-match objects and reproduce plots. Human explanations make the catalog accessible to students and researchers in different languages. The two needs are related, but they are not identical.
If a translator changes a column’s display label from “Flux density” to a natural equivalent in Mandarin, that may be useful. If the same operation changes the stored unit, the UCD, or the values themselves without a documented conversion, the resulting table can cease to represent the same experiment or observation. The fundamental rule is to separate presentation from scientific semantics.
Before touching a dataset, create a field inventory with original ID, name, datatype, arraysize, unit, ucd, ref, descriptive text and known consumers. Record whether a field is measured, derived, categorical or purely explanatory. Translation rights should be scoped to the fields that are actually natural-language content, not every sequence of characters in the XML file.
VOTable 1.5 is an XML scientific-data standard
VOTable is designed for tabular data used in the Virtual Observatory ecosystem. A VOTable document contains resources, tables and field declarations, followed by data in supported serializations. The schema defines where elements can appear and which attributes they accept. Language localization must respect that structure rather than introducing arbitrary new tags because they resemble familiar web markup.
In version 1.5, the root uses version="1.5" while its VOTable element namespace remains http://www.ivoa.net/xml/VOTable/v1.3. This apparent difference is deliberate, according to the Recommendation. Changing the namespace string to a guessed v1.5 would be a technical mistake even though it might appear more intuitive to a translator.
The specification’s discussion of potential future extensions is also important. A later appendix illustrates ideas such as multiple descriptions or language-aware metadata that are not simply ordinary conforming features of the adopted standard. Do not insert extra <DESCRIPTION> variants or an unapproved xml:lang attribute and assume every 1.5 validator will accept the result.
Field names and UCDs are not synonyms
The name and ID attributes help identify fields and references within the table or its user interfaces. A UCD expresses standardized semantic information about the physical quantity. The associated unit tells a consumer how to interpret numerical scale. A natural-language description should explain these, but it does not replace them.
An English heading such as “Brightness” can be ambiguous: it may refer to a magnitude, flux or flux density, each with different numerical interpretation. A translation that chooses a common word equivalent to “light level” can hide these distinctions. The correct explanation must be grounded in the UCD, unit and supporting scientific documentation, not merely the familiar source-language heading.
Test with two nearby fields, such as magnitude and flux density, shown in the same translated table. A student should be able to state what each column measures and what its unit means. The machine-readable metadata must still allow an independent analysis program to interpret both columns without using the translated display text.
How to build a safe translation workflow
Pass one: determine which text is truly natural-language metadata
Review DESCRIPTION content for the table and each field, plus any project-provided catalogue notes. Distinguish prose descriptions from controlled terms, identifiers, units, mathematical expressions and references. A field name may be concise and human-readable but still serve as a stable key in a plotting script or query definition.
If existing tools depend on a particular column name, renaming it is a schema migration rather than a mere translation. Prefer a localized label stored in an external companion catalogue or presented by the user interface. Document the mapping between the unmodified field identity and each target-language explanation.
Pass two: read the underlying physical quantity
Before translating a field description, inspect its data type, unit, UCD and any linked coordinate or time reference. Ask the catalog scientist what quantity is represented, how it was derived, and which qualifications are important. Technical symbols may appear ordinary to a general translator but can encode measurement assumptions and analysis conventions.
Write a neutral-language definition of each field before selecting target words. This definition should include the quantity, reference frame or model assumption, known constraints and whether the value is a measurement or derived estimate. A translator can then choose natural terms without manufacturing new scientific facts.
Pass three: preserve a clean original table
Keep the source VOTable or another authoritative snapshot intact. If an organization chooses to edit an actual permitted descriptive element within the XML, validate the result against the applicable VOTable schema and confirm downstream application compatibility. For several languages, a companion display dictionary is often simpler and safer than rewriting scientific files for every locale.
A companion dictionary might map a stable field ID to en and zh-Hans labels, each with clear ownership and versioning. This dictionary is an application artifact, not a claim that VOTable 1.5 natively provides that exact multilingual map. Keeping the boundary explicit allows the original catalog to remain reproducible.
Pass four: validate numeric and structural invariants
Run a before-and-after comparison that checks table count, field count, field order, datatypes, arraysizes, units, UCDs, references, row counts and values. Where the data uses binary serialization, compare parsed scientific values rather than assuming that textual appearance tells the full story. Confirm that no converted character encoding or XML edit has changed the data payload.
Use a representative file with missing values, non-ASCII descriptions, exponent notation and more than one resource. A localization feature should not break on the first realistic edge case. A successful structural test means the same analysis program can load the file and obtain equivalent data before and after any permitted descriptive change.
Pass five: test scientific understanding with real tasks
Ask an astronomy reader to identify the coordinate system, select a field in the correct unit, describe the meaning of an uncertainty column and explain the difference between missing and zero. Compare responses in English and the target language. If a reader makes a materially different scientific conclusion after translation, the wording needs repair.
A useful test is to reproduce a small plot from each interface language. The same numeric column and unit transformation should produce the same plotted points, not merely similar axis labels. This checks the application mapping, data binding and scientific interpretation together.
Pass six: preserve provenance and review limitations
A translated description should identify whether it comes from the catalog producer, the host application or a third-party educator. Scientific articles may refer to instrument calibration and data-release versions that change over time. When a translation is updated, keep the original source, reviewer and revision date so that later users can understand why the explanatory label changed.
Do not call a translation officially endorsed by the observatory unless that approval exists. A well-written educational gloss may still be a separate interpretation. Mark the difference especially when a catalogue is used for peer-reviewed work, citizen-science classification or a student’s assessment of published research.
Worked VOTable XML example
This compact teaching sample uses the VOTable 1.5 namespace and version combination established by the IVOA Recommendation. It records one illustrative numeric flux-density value and a descriptive element. It is not an actual astronomical observation, and XML parsing alone does not prove schema conformance or scientific plausibility.
<?xml version="1.0" encoding="UTF-8"?>
<VOTABLE version="1.5" xmlns="http://www.ivoa.net/xml/VOTable/v1.3">
<RESOURCE name="exampleSurvey">
<TABLE name="fluxTable">
<DESCRIPTION>Illustrative flux-density measurement.</DESCRIPTION>
<FIELD ID="flux" name="flux" datatype="double"
unit="Jy" ucd="phot.flux.density">
<DESCRIPTION>Flux density in janskys.</DESCRIPTION>
</FIELD>
<DATA>
<TABLEDATA>
<TR><TD>1.25</TD></TR>
</TABLEDATA>
</DATA>
</TABLE>
</RESOURCE>
</VOTABLE>A suitable companion display may show a Chinese explanation of the field while retaining ID="flux", unit="Jy", the UCD and numerical value. Do not insert a new <LABEL lang="zh"> element inside <FIELD> merely because such a structure would be convenient; the VOTable schema has its own defined content model. A separately versioned interface catalogue is a more transparent approach.
Suppose a user wants flux density expressed in another unit. That is a data transformation with dimensional checks, rounding policy and provenance, not a translation. The interface can translate the word describing the unit or show a validated conversion, but must identify the conversion explicitly and preserve the original measurement. Otherwise two locale views can appear to report different observations.
Scientific translation cases and repair methods
1. Right ascension is labeled as ordinary time
A catalog includes right ascension expressed in degrees. A translator chooses a phrase meaning “time of day” because the English term contains “ascension” and the application sometimes displays coordinates in hours. A reader may mistake angular position for a clock value and compare it against the wrong axis.
Inspect the field’s UCD, unit and coordinate reference. Right ascension is an angular coordinate in a specified celestial coordinate system, even when an interface uses hour-like notation. Its physical and mathematical meaning should be explained before selecting local terminology.
Use the established astronomy term in the target language and state the displayed unit. Test the same position represented in degrees and in validated hour-angle formatting. A successful reader knows that a change of representation does not change the actual celestial position or convert the value into observation time.
2. Declination is confused with downward movement
The word “declination” can resemble an everyday expression for decrease or decline. A translation turns a column of celestial declination into “falling angle,” inviting students to interpret it as an object’s motion or change over time. The data actually records a coordinate component, not a measured trajectory.
Check the coordinate system and the UCD before choosing the term. Positive and negative declination need a clear relationship to the celestial equator or relevant reference convention. A single value does not imply that a source is moving north or south.
Translate the coordinate concept, not the English word’s ordinary connotations. Test two sources with opposite declination signs and ask readers to locate them conceptually. The value and sign must remain unchanged, and the explanation should not manufacture a velocity or temporal trend.
3. A `Jy` unit is expanded as a person's name
A field uses unit="Jy", an astronomy unit symbol, and the translator replaces it in the raw table with a localized spelling of the word “jansky.” The result is friendlier to some readers but may no longer satisfy a consumer expecting the standardized unit syntax or established unit handling.
Treat unit attributes as machine-readable scientific symbols, not prose labels. Inspect their standard interpretation and the receiving application’s unit parser. Translating the unit name for a tooltip is different from rewriting the attribute’s value, and a display spelling should not become the authoritative machine unit.
Retain Jy in the data and provide a target-language explanatory label outside the unit attribute. Test a plotting program that reads the field and generates a numerical axis. If the application parses the original and translated file differently, the localization has crossed the scientific data boundary.
4. Magnitude becomes brightness on a linear scale
A student sees a translated field titled simply “Brightness” and assumes that larger numerical values mean a brighter object. If the underlying quantity is an astronomical magnitude, the scale may behave oppositely to that intuition and can use different photometric systems. The label is therefore scientifically incomplete.
Read the UCD, unit conventions and catalog documentation. Determine whether the field represents apparent magnitude, absolute magnitude, flux or another derived quantity, and whether a bandpass or calibration is specified. Translation must not flatten these distinctions into a familiar but misleading everyday adjective.
Use a scientifically precise localized name and a brief explanatory note where the reader needs it. Test two sample values and ask which object is brighter under the stated magnitude convention. The user’s answer should match the actual scale, not the intuition induced by an overly generic translation.
5. Measurement uncertainty is translated as a confidence score
A numeric uncertainty column follows a measured flux value. The display calls it “confidence,” and students interpret 0.2 as 20% probability of correctness. A scientific uncertainty may instead denote an estimated standard deviation or another error measure in the original quantity’s units.
Inspect the uncertainty model and its unit. Look for UCDs and documentation indicating statistical versus systematic errors, and whether the value is absolute or relative. A table alone may not state all assumptions; the relevant survey paper or data release notes can be necessary to interpret the column honestly.
Translate the source’s actual statistical meaning and state limits where required. Test a case where uncertainty and value have the same dimensional unit. The user should not infer a confidence percentage unless a probability interpretation is explicitly supported by the source method.
6. Column `ID` changes when a title is localized
A plotting script references a field by its stable ID. A translation plugin renames that identifier to a Chinese phrase because the column should appear translated in the interface. The XML may remain well-formed but internal references or saved analysis scripts can fail or select an unexpected field.
Separate the field’s machine identity from its human display label. Check any references and application queries before modifying structural attributes. Renaming a stable field is a coordinated data-model migration with consumer impact, not a prerequisite for showing a localized table heading.
Keep the identifier and bind it to an application-level localized label. Run the same saved plot and data extraction in English and Chinese. The selected field and numeric values must be identical even though the visible wording changes, and a copied technical identifier should remain usable.
7. A field `name` is treated as disposable prose
A FIELD has a readable name that a research group uses in queries and visualization scripts. A translator edits the name attribute for clarity, unaware that downstream tools treat it as an expected key. The table looks better in a generic viewer but breaks a reproducible analysis notebook.
Inventory client dependencies on names and IDs separately. The VOTable model permits different forms of metadata, but application contracts can impose stricter expectations than the XML schema alone. A field may be technically editable while changing it is operationally unsafe for the data producer’s published interface.
Use a companion presentation mapping or negotiate a versioned schema change with the owner. Test named-column selection in a representative downstream application. Include one client that matches by name and another that matches by ID so the review catches hidden assumptions in real analysis workflows.
8. UCD vocabulary is translated into ordinary words
A field’s ucd encodes the kind of astronomical quantity. A translator replaces a UCD component with a local-language equivalent, thinking it is a label. Data discovery software no longer recognizes the controlled semantic expression, potentially excluding the column from searches for a particular physical concept.
Treat UCD content as vocabulary syntax owned by the relevant standards, not free text. Translate its explanatory definition for readers if needed, but preserve the machine expression. Check combinations and semicolon-separated structure according to the applicable UCD rules instead of translating individual fragments.
Restore the original UCD and put the explanation in a sanctioned descriptive channel. Test automated discovery before and after localization: a query for flux-density columns should return the same field set. The goal is a clearer display without losing interoperability across astronomical datasets.
9. XML namespace is guessed from the version number
A developer sees version="1.5" and changes the namespace URI to end in v1.5. The change appears symmetrical but contradicts the VOTable 1.5 Recommendation, which uses the established v1.3 namespace. An XML parser may read the document, yet a VOTable-aware tool can reject the unexpected namespace.
Compare the root attributes against the official VOTable 1.5 example and schema. Avoid inferring namespaces from naming patterns or from generic XML intuition. The version and namespace are separate technical declarations, and their precise pairing is part of the format’s compatibility story.
Restore the specified namespace and version combination. Parse the XML, then validate it with a suitable VOTable schema or library. Test import through a real astronomical viewer, because a generic XML parse can succeed even when the scientific-data consumer refuses the file.
10. Several translated descriptions are inserted as sibling elements
A team wants English, Spanish and Chinese versions of a field description, so it adds three <DESCRIPTION> elements with guessed lang attributes. The output looks like a sensible multilingual document, but the adopted VOTable schema may allow only one description in that position and does not necessarily recognize the added attributes.
Distinguish normative features from ideas discussed in the specification’s extension appendix. An illustrative future proposal is not permission to use a new construction in production VOTable 1.5. Check element cardinality and allowed attributes with the actual schema before inventing a multilingual markup convention.
Store translations in a companion catalog keyed to stable source fields, or use a rigorously documented supported extension approved by all consumers. Test a strict validator and a real client, not just a forgiving XML editor. Maintain the original description so users can reconstruct what the catalog producer intended.
11. A null value is converted into zero
A translator replaces blank cells with a localized word for “none,” but the export code converts the missing numeric value to 0. Scientists then treat nondetections or unavailable measurements as actual zero-valued observations. This can distort averages, fits and cross-matches even though the visible table seems more readable.
Inspect the format’s handling of missing values and the specific field datatype or serialization. The meaning of a blank cell, a marker, an unavailable quantity and a measured zero must not be guessed from display conventions alone. Review data-release notes when survey-specific flags are involved.
Translate the interface’s missing-value label without modifying underlying values. Test a statistical calculation with a missing entry and a true zero, and confirm the two cases remain distinct. A useful localized display helps readers understand absence of evidence rather than introducing new scientific evidence.
12. Exponent notation is rewritten as a decimal sentence
A flux or distance column contains scientific notation. A localization routine treats 1.2E-6 as text and expands it with locale punctuation, producing a string a numeric parser cannot read. A viewer may display the result, but an analysis program may reject the row or silently treat it as missing.
Check the field datatype and the chosen VOTable serialization. Numeric lexical forms have format-specific rules that differ from a human sentence or localized screen display. A translator should not modify scientific-number representations in stored data merely to match local decimal conventions.
Keep numeric values typed and format them for human display at the interface boundary. Test conversion back to the same quantity, including negative exponents and boundary magnitudes. If the product supports copying formatted values, make clear whether the copy action returns a display string or a machine-readable number.
13. Array sizes are mistaken for presentation width
A FIELD declaration uses arraysize to describe the shape or size of a data value. The localization team assumes it is the maximum visible character count and changes the attribute to fit a longer translated label. This can break the declared structure of actual array or string values.
Read arraysize in the context of VOTable data modeling, not user-interface typography. The appropriate solution to long labels is a responsive column heading or an accessible detail view, not an alteration to the field’s data-shape metadata. Test with real values near the declared array-size boundary.
Restore the source field declaration and improve layout separately. Validate serialized data against the original field shape and rerun representative readers. Include both a short and long translation in the UI test, because visual truncation and data-schema compatibility need separate fixes.
14. A coordinate frame is omitted from the translation
A catalog contains coordinate values defined in a particular reference frame. The translated description simply says “sky position,” and readers combine the values with another catalog in a different frame without transformation. The resulting positional discrepancy looks like a real astronomical offset rather than a metadata interpretation error.
Inspect COOSYS or the referenced coordinate metadata and any relevant external survey documentation. Coordinate values are meaningful within a frame and epoch convention where specified. A translation should carry essential frame assumptions rather than reducing technical context to a general phrase that happens to be easy to understand.
Preserve frame identifiers and make the necessary explanation visible in the target language. Test a cross-match requiring frame-aware handling, and confirm the application uses the original metadata, not the translated sentence, to decide how values should be compared. Translation does not perform coordinate transformations.
15. Observation time is confused with catalogue release date
A table contains time-related measurements and an informational description of when the data release was published. A target-language summary uses one word for both, causing students to plot release dates as if they were observation times. Historical trends are then distorted by administrative metadata.
Review TIMESYS where relevant, the meaning of time fields and the source provenance of release notes. A time coordinate and a publication date can both be valid dates while playing very different analytical roles. Do not infer the correct role from the appearance of an ISO-like string alone.
Use distinct translated labels for observation, reference epoch, processing and publication time. Test a time-series plot and a data-release citation separately. A correct localization lets a reader choose the intended time dimension without changing the time value or the scientific interpretation of the sample.
16. Angular units are mixed in an axis label
A translated plot labels the horizontal axis “hours” because right ascension is often expressed in hour angles, while the VOTable field reports degrees. The plotted points remain numerically in degrees. A student reading the chart believes the sky positions are fifteen times different in angular interpretation.
Compare the plot’s selected field, unit attribute and any actual conversion function. The mere possibility of representing right ascension in hours does not mean the current data has been converted. An axis label must describe the values actually plotted, not a common alternative convention.
Keep the degree label unless a validated transformation has been applied. If the product offers hour-angle display, test that the values and labels change together and that a switch back recovers the originals. Document rounding and wraparound behavior as numerical concerns, not linguistic ones.
17. Derived values are presented as direct observations
A table contains estimated distance values computed from another measurement under stated assumptions. The translation calls them “observed distance,” implying the telescope directly measured the distance. This changes the epistemic status of the number and can mislead a reader about uncertainty and model dependence.
Look for description language about estimates, calibrations and assumed constants. Determine what is measured directly and what is derived by a model or calculation. Terms such as inferred, estimated and measured should not be treated as interchangeable stylistic choices in a scientific metadata translation.
Restore the distinction and include the necessary assumption in the localized help text. Test a case where the assumed model changes the estimate. The reader should understand why a different derived distance can appear without believing the instrument itself observed two contradictory physical distances.
18. A classification label is treated as a new object type
A field holds a source classification produced by one survey. A translator substitutes a broad popular-science category that does not match the survey’s technical classification codes. Users then count objects under the wrong classes or compare categories that were not equivalent across datasets.
Inspect the class-code vocabulary, survey method and any uncertainty flags. A row’s classification may be provisional, probabilistic or release-specific. The target wording should preserve that status and not suggest a universally accepted astrophysical taxonomy when the source only offers a model’s label.
Translate the category explanation using a controlled mapping while retaining the underlying code and provenance. Test two closely related classes and a low-confidence example. The localized table should support the same selection set and clearly indicate where a classification remains uncertain.
19. A comment in the data is mistaken for a physical fact
A catalog producer writes “possible blend near field edge” in a descriptive note. The translated table states “two stars confirmed,” turning a cautious quality note into an asserted astronomical conclusion. Readers may cite the transformed statement as evidence for a multiple source that the survey did not establish.
Read hedging language, quality flags and nearby metadata together. Scientific annotation can express uncertainty, observational limitations or suspected artifacts. A translator must preserve these qualifications, including whether the note is about data quality rather than the physical properties of the object.
Translate the uncertainty explicitly and separate observer commentary from validated measurements. Test a reader asked what is established versus suspected. The translation is acceptable only when it communicates the same evidence boundary as the source, even if a shorter confident sentence would fit better in a column tooltip.
20. Links to supporting data become localized destinations
A LINK or companion resource points to a calibration table or survey documentation. A translation layer rewrites the URL path to resemble a localized title, but the target file does not exist. The displayed text is readable, while the reference chain needed to reproduce the measurement is broken.
Separate link labels from URI targets and check whether the publisher provides genuine alternate-language resources. Do not derive new URLs by translating path segments. Supporting documentation may be language-specific, but scientific data links and identifier targets must remain grounded in actual resources.
Keep the original target unless a verified locale alternate exists, and record that alternate explicitly. Test all links from the localized interface, including references embedded inside explanatory text. A successful translation should improve access to provenance, not sever it for readers using another language.
21. A table order change appears as a translation
The user interface sorts the table by a localized string label instead of the underlying numeric field. In one locale, thousands separators or digit formatting change the sort order, creating a misleading ranking of flux values. The astronomer may interpret the list as a real scientific ordering.
Check whether sorting operates on typed data or rendered strings. Scientific values should retain their numeric comparison semantics regardless of the interface language. Similar mistakes affect negative values, nulls and exponent notation, where lexicographic ordering is particularly deceptive.
Sort by the original typed value and localize only the displayed rendering. Test negative, zero, missing and large exponent values in several locales. The row order should remain equivalent for the same sort key, with an explicit policy for missing data rather than a language-dependent accident.
22. Unicode description editing corrupts XML entities
A translator inserts an ampersand or angle bracket into a scientific description without proper XML escaping. A generic viewer may hide the problem until a strict parser rejects the entire document. The failure occurs in prose, but its consequence is loss of access to the whole dataset.
Use an XML-aware editor and serializer rather than raw string replacement. Check character encoding, entity escaping and legal XML code points. Unicode text itself can be supported, but the surrounding XML syntax has strict rules. Do not treat punctuation normalization as a harmless formatting operation when it is performed on raw markup.
Validate well-formedness and schema compatibility after permitted descriptive edits. Test a target-language sentence with special characters and non-Latin text. If multiple locales are needed, prefer a separate display catalog that leaves the scientific XML untouched while allowing richer multilingual typography.
23. A single field description is reused across incompatible tables
Two catalogs use the field name flux but represent different wavelength bands and calibration assumptions. A translation-memory system reuses the same localized description because the field names match. Users may compare the columns as though their numerical values were directly interchangeable.
Include table, instrument, band, unit, UCD and release context in terminology decisions. A concise field name alone is not enough to determine the scientific meaning. The relevant unit and metadata may establish a distinction that the source-language labels do not reveal.
Store translation mappings with source-specific scope and require a scientific reviewer for reuse across projects. Test two flux fields in different resources and ask whether direct comparison is justified. A useful interface should make context visible rather than hiding it behind a universal translation-memory match.
24. A binary serialization is edited as if it were a text table
A VOTable document uses BINARY2 or another supported data serialization. A localization process searches raw bytes for recognizable English strings and changes them, damaging encoded values or offsets. The document may still carry a plausible XML wrapper while the scientific data payload is corrupted.
Inspect the DATA serialization and use a proper VOTable reader. Not every value appears as a visible XML <TD> cell. Translators should operate on permitted descriptive metadata or an external UI layer, not blindly replace byte sequences inside embedded or linked data streams.
Compare parsed values, field declarations and row counts before and after any processing. Test a binary example in addition to a simple TABLEDATA fixture. A successful localization preserves numerical content regardless of how the VOTable stores it, including the cases that cannot be reviewed by opening the file in a text editor.
25. An export drops the catalog producer's original description
A viewer downloads a “translated VOTable” that has replaced the source producer’s scientific description with an editorial explanation. Later researchers cannot distinguish the original catalog author’s wording from the translation team’s interpretation, making it difficult to reproduce how a field was defined in the source release.
Establish which metadata belongs to the data producer and which belongs to the local teaching interface. If the source file is rewritten, provenance and license requirements may matter. A companion catalog is often safer because it can preserve both the original wording and the revised educational explanation.
Export an unmodified source where possible, and keep translated annotations separate or clearly attributed in a supported representation. Test by reopening the downloaded file in an independent program and comparing its descriptive and numerical metadata against the archival source. Avoid calling an editorial gloss an official replacement unless authorized.
26. A student interprets a missing measurement as a nondetection
An astronomy lesson displays a blank field with the translated word “not detected.” The source actually indicates that the measurement was not available for that object, which may be due to field coverage rather than lack of emitted light. The localization turns missing evidence into evidence of a physical absence.
Review data-quality flags, null conventions and survey coverage statements. “Not measured,” “not available” and “not detected above a stated threshold” are scientifically different propositions. The appropriate wording depends on what the catalog documents, not a generic missing-value phrase.
Translate the specific reason if one is represented; otherwise use a neutral missing-data label. Test a source outside instrument coverage and a measured nondetection with an upper limit. The learner should be able to distinguish both cases and avoid false conclusions about the source’s physical properties.
Changed-context transfer exercise: multilingual astronomical survey workshop
A school astronomy club downloads a VOTable with right ascension in degrees, declination, an apparent magnitude field, flux density and one missing value. Teachers want Chinese and Tamil explanations in an interactive table. A volunteer proposes changing name, unit and ucd attributes into local-language text and inserting multiple <DESCRIPTION lang="..."> elements. Decide which parts of the plan are safe.
The correct answer is to retain the VOTable’s technical field declarations, numerical values and coordinate metadata. Build a companion display dictionary keyed to stable field identities for the Chinese and Tamil labels, and use an interface to show explanatory notes. The source VOTable should retain the recommended version and namespace pairing, and any permitted edits should be validated rather than relying on invented multilingual markup.
Students should test a sky-position plot, compare magnitude interpretations and examine the missing value without converting it into zero. They should also verify that the same row selection and scientific values appear in both interface languages. A translated explanation can introduce astronomy vocabulary without changing the data or silently converting degrees, magnitudes or flux units.
Practical review checklist
- Keep VOTable version, XML namespace, resource/table structure and data serializations valid.
- Preserve field identities,
datatype,arraysize,unit, UCD, references and numeric content. - Review
COOSYS,TIMESYS, uncertainty and scientific data-release context. - Distinguish natural-language descriptions from standardized metadata and controlled terms.
- Use a companion translation catalogue where native multilingual markup is unsupported.
- Test nulls, negative numbers, exponent notation, right-to-left text and Unicode escaping.
- Reproduce one analysis or plot in every supported language and preserve provenance.
Limits and uncertainty that remain after translation
VOTable localization does not determine whether a source measurement is calibrated correctly, whether a classification model is valid, or whether a plotted association is scientifically justified. Those require scientific methods, survey documentation and reproducible analysis. A well-translated table can help readers recognize uncertainty; it cannot remove uncertainty from the measurement itself.
The sample above is checked for XML well-formedness and selected field invariants, not full VOTable 1.5 conformance. A production system must use an appropriate schema validator, VOTable reader and downstream astronomy tools, especially for less familiar serializations and metadata configurations. Nor should proposals described in a non-normative extension appendix be represented as already-supported standards features.
Summary: translate explanations, not the universe being measured
A fast and reliable VOTable translation helps a multilingual reader understand what each column means while preserving everything that makes the data scientifically reusable. Treat units, coordinate frames, data types, UCDs and numeric values as part of the measurement contract. Keep descriptions precise, validation independent and original provenance recoverable. Then a change of language remains a change of presentation rather than a change of the astronomical result.
Frequently asked questions
1. Can VOTable 1.5 store any number of translated `<DESCRIPTION>` elements?
Do not assume so. Check the adopted schema; multilingual extension ideas described in a non-normative appendix are not automatically conforming VOTable features.
2. Why does a VOTable 1.5 file use a namespace ending in `v1.3`?
That is the namespace specified by the VOTable 1.5 Recommendation; version and namespace strings are not required to have matching numerals.
3. Is the field's `name` safe to translate?
It may be used as a key by tools. A separately localized display label is often safer than changing the source field attribute.
4. May I translate the `ucd` attribute into the local language?
No. Preserve the controlled scientific semantic expression and translate its explanation separately.
5. What happens if a numeric column is formatted with local punctuation?
The display may change, but the stored numeric value and its parsing semantics must remain stable.
6. Can a missing value be displayed as zero?
Not unless the underlying scientific data actually represents zero. Missingness and a measured zero are distinct.
7. Does translating a unit symbol also convert the measurement?
No. Unit conversion requires a defined numerical transformation, not a vocabulary substitution.
8. How should coordinate frames be explained to students?
Use clear localized text while retaining frame identifiers and documenting any actual coordinate transformation separately.
9. Is well-formed XML sufficient for VOTable 1.5 compatibility?
No. The file still needs schema-level and application-level validation, including scientific value checks.
10. What is the best practical test?
Reproduce the same query, table selection or plot in the source and target-language interfaces and compare the underlying values.
Authoritative sources
- IVOA VOTable 1.5 Recommendation — 16 January 2025
- IVOA VOTable 1.5 Full Recommendation
- IVOA UCD1+ Recommendation
- IVOA VOUnits Recommendation