SciMetrics
Back to Docs
Natural-language user guide

100 Ways to Work with SciMetrics

From data import to AI topics, PMRC, author mapping, and cross-fusion, find a similar task and start with one natural-language request.

100 examples
001
Getting started

Check the SciMetrics connection for the first time

Say this

Check whether SciMetrics is connected. Read only the current state and available capabilities, do not change any data, and explain in plain language what I can do now.

View the full conversation path

What SciMetrics will do

The assistant connects through SciM MCP and reads the application, dataset, and available task state.

Continue with

  • Am I using desktop mode or background mode?
  • Recommend three read-only tasks that are suitable for a first try.

Another way to ask

Help me confirm that SciMetrics is ready to accept my requests.

What you will get

A connection summary and a list of task types that are ready to use.

002
Getting started

Understand the open dataset

Say this

Read the dataset currently open in SciMetrics and tell me which tables it contains, how many records each has, and which years it covers. Give me an overview only; do not run an analysis.

View the full conversation path

What SciMetrics will do

SciMetrics reads the dataset structure, record counts, and basic time coverage.

Continue with

  • Which tables are most useful for a network analysis?
  • Point out important fields that are clearly missing or empty.

Another way to ask

Give me a guided overview of the dataset that is open now.

What you will get

A structured overview of the current dataset without creating analysis results.

003
Getting started

Ask for recommendations when you do not know where to start

Say this

I am not sure which analysis to run. Based on the current data type, fields, and years, recommend five worthwhile SciMetrics tasks and explain which question each task can answer. Do not execute anything yet.

View the full conversation path

What SciMetrics will do

The assistant checks the live dataset and available capabilities before organizing suitable research tasks.

Continue with

  • Put the two easiest tasks first.
  • Keep only suggestions related to topic evolution.

Another way to ask

Look at the current data and tell me which research questions SciMetrics is best suited to answer.

What you will get

Task recommendations matched to the current data rather than a generic feature list.

004
Getting started

Request a plan before execution

Say this

I want to study institutional collaboration and topic change in the current data. Split the work into no more than four stages, list the input, action, and output for each stage, and wait for my confirmation before starting.

View the full conversation path

What SciMetrics will do

The assistant turns the combined goal into verifiable stages and pauses before writes or long-running work.

Continue with

  • Make stage one a data-quality review only.
  • Add a stopping condition for every stage.

Another way to ask

Give me a roadmap for institutional collaboration and topic change, but do not run it yet.

What you will get

A staged workflow that can be reviewed and approved one step at a time.

005
Getting started

Limit a request to read-only inspection

Say this

Read only the current data and window state. Check record counts, the selected scope, and the current analysis type. Do not save, overwrite, merge, or delete anything.

View the full conversation path

What SciMetrics will do

SciMetrics performs state and data reads only and avoids operations that change the dataset.

Continue with

  • Turn the inspection into a short checklist.
  • Tell me which possible next actions would modify data.

Another way to ask

Run a completely read-only health check first.

What you will get

A checklist of the current data and interface state with no data changes.

006
Getting started

Ask only for information that is truly missing

Say this

Help me analyze the annual trend in the current data. Inspect what is already known and ask only when a missing decision would affect the result. Do not ask me for command names or API parameters.

View the full conversation path

What SciMetrics will do

The assistant reuses the current state and asks only about research choices such as years, entities, or output.

Continue with

  • Use all checked documents by default.
  • Show the result on screen first; no file export is needed.

Another way to ask

Run an annual trend analysis and ask me only for essential decisions.

What you will get

An annual-trend plan or view with unnecessary technical questions removed.

007
Getting started

Break a large request into manageable stages

Say this

This task includes import, cleaning, AI topic recognition, network analysis, and export. Divide it into separately verifiable stages. Run only import and quality checks now; list the remaining stages without executing them.

View the full conversation path

What SciMetrics will do

The assistant establishes a stage order, runs only the first verifiable stage, and preserves the later plan.

Continue with

  • Move to cleaning only after stage one has no unresolved anomaly.
  • Report the actual processed count after every stage.

Another way to ask

Do not do everything at once; import the data and check its quality first.

What you will get

A verified first-stage result and a clear list of remaining tasks.

008
Getting started

Continue a task after an interruption

Say this

The previous SciMetrics task was interrupted. Read its status, completed steps, and existing outputs first. Tell me where it is safe to continue, and do not rerun every step automatically.

View the full conversation path

What SciMetrics will do

The assistant checks the terminal state, progress, and outputs to separate completed work from remaining work.

Continue with

  • Retry only the failed batches.
  • Before continuing, confirm that existing outputs will not be overwritten.

Another way to ask

Help me resume the interrupted work without starting over.

What you will get

A recovery recommendation and the smallest necessary retry scope.

009
Getting started

Ask the assistant to explain a failure

Say this

The last operation failed. Stop repeating it. First determine whether the issue is the connection, application startup, data conditions, task selection, or file writing, and give me the smallest repair step.

View the full conversation path

What SciMetrics will do

The assistant reads the error and state, classifies the failure, and avoids another write before the cause is addressed.

Continue with

  • Was any data partially changed?
  • After the repair, run a read-only verification first.

Another way to ask

Do not try again yet; explain what failed and how to repair it with the smallest change.

What you will get

A plain-language error category, impact assessment, and repair order.

010
Getting started

Keep a reproducible operation record

Say this

After the current analysis, prepare an operation record containing the data used, filtering scope, steps actually completed, main settings, and output files. Do not include accounts, keys, or license contents.

View the full conversation path

What SciMetrics will do

The assistant summarizes inputs, scope, actions, and actual files after the task has completed.

Continue with

  • List incomplete items and human-review decisions separately.
  • Also create a shorter version suitable for a research methods section.

Another way to ask

When this analysis ends, leave a process note that another person can review.

What you will get

A reproducible analysis record without sensitive information.

011
Data import and preparation

Import Web of Science records

Say this

Import D:\Research\wos-records.txt into SciMetrics. Identify the appropriate import method first, then check the document count, year range, and whether author and affiliation data are present.

View the full conversation path

What SciMetrics will do

SciMetrics selects a matching import filter, creates the dataset, and performs basic completeness checks.

Continue with

  • List records that were skipped or could not be recognized during import.
  • Save this as a new SciMetrics dataset.

Another way to ask

Import this WoS file and confirm that its main fields are usable.

What you will get

A reusable dataset and an import-quality overview.

012
Data import and preparation

Detect the import format when it is unknown

Say this

I do not know which import format this file needs. Inspect D:\Data\records.txt and the available import methods, recommend the best match with a reason, and wait for confirmation before importing.

View the full conversation path

What SciMetrics will do

The assistant compares file characteristics with currently supported SciMetrics filters without asking for a technical filter name.

Continue with

  • Preview a few records that can be recognized first.
  • If two options are possible, explain the difference.

Another way to ask

Figure out how this literature file should be imported instead of asking me to guess a format name.

What you will get

An import recommendation followed by a confirmed dataset.

013
Data import and preparation

See which tables are in the dataset

Say this

List all tables in the current dataset and explain in one sentence what the main Document, Author, Affiliation, Keyword, and related tables contain.

View the full conversation path

What SciMetrics will do

SciMetrics reads the table catalog and table-level summaries and presents them without technical schema language.

Continue with

  • Show the record count for every main table.
  • Tell me which tables are related to each other.

Another way to ask

Introduce the contents of this dataset table by table.

What you will get

A table catalog with purpose and size summaries.

014
Data import and preparation

Keep a specified publication period in scope

Say this

Select records published from 2020 through 2025 in the current documents. Tell me how many match first, and do not delete records from other years.

View the full conversation path

What SciMetrics will do

SciMetrics filters and updates the selection scope without deleting out-of-range records.

Continue with

  • Use these documents as the current analysis scope.
  • List the selected document count by year.

Another way to ask

Use only 2020-2025 for the next analysis while keeping the original data intact.

What you will get

A year-limited analysis selection with the full dataset preserved.

015
Data import and preparation

Find and select documents by keyword

Say this

Search titles, abstracts, and keywords in the current documents for solid-state battery. Show the match count and a few representative records, then mark the matching documents as selected.

View the full conversation path

What SciMetrics will do

SciMetrics searches the specified text fields, previews matches, and updates document selection.

Continue with

  • Include lithium-metal battery in the search condition.
  • Keep only records whose abstracts clearly discuss the topic.

Another way to ask

Find and select documents related to solid-state batteries in the current data.

What you will get

Reviewable matches and a selection scope ready for later analysis.

016
Data import and preparation

Select all documents again

Say this

Tell me how many documents are currently selected, then clear the current filter and select all documents again. Do not change any field values.

View the full conversation path

What SciMetrics will do

SciMetrics reads the current selection count and restores the analysis scope to all documents.

Continue with

  • Confirm that the total and selected document counts now match.
  • Next, filter them by document type.

Another way to ask

Clear the previous document filter and include every record again.

What you will get

A full-dataset document selection.

017
Data import and preparation

Check missing key fields

Say this

Check year, title, abstract, author, affiliation, and keyword fields in the current documents. Report the missing count and percentage for each. Inspect only; do not fill values automatically.

View the full conversation path

What SciMetrics will do

SciMetrics calculates missingness for important fields while leaving the source data unchanged.

Continue with

  • Sort the fields by severity of missingness.
  • Find documents missing both abstract and keywords.

Another way to ask

Run a completeness check on the key literature fields.

What you will get

A missing-field summary and a scope for priority review.

018
Data import and preparation

Preview institution-name merge suggestions

Say this

Find name variants in the affiliation table that may represent the same institution. Show the original name, proposed standard name, affected record count, and reason first. Do not merge yet.

View the full conversation path

What SciMetrics will do

SciMetrics inspects similar affiliation names and prepares normalization suggestions for human confirmation.

Continue with

  • Show only high-confidence suggestions.
  • Exclude similarly named institutions located in different cities.

Another way to ask

Find institution-name variants and let me review the merge suggestions before applying them.

What you will get

A reviewable list of candidate institution-name normalizations.

019
Data import and preparation

Clean only checked records

Say this

Process only the checked rows in the current table. Normalize extra spaces and obvious capitalization differences in institution names. Preview the affected rows first and save only after confirmation.

View the full conversation path

What SciMetrics will do

SciMetrics limits the operation to checked rows, prepares a change preview, and updates safely normalizable text after approval.

Continue with

  • Do not change institution IDs.
  • After saving, tell me how many rows were actually updated.

Another way to ask

Clean only the institution names I checked, with a preview before any change.

What you will get

A confirmed local name-cleaning result and its actual update count.

020
Data import and preparation

Save and reopen a dataset

Say this

Save the current dataset to D:\Research\cleaned-project.scim, close the current data, reopen the saved file, and confirm that the main tables and record counts remain consistent.

View the full conversation path

What SciMetrics will do

SciMetrics saves the dataset, reloads it, and compares key table counts.

Continue with

  • Tell me the actual size of the saved file.
  • After reopening, check the current selection scope.

Another way to ask

Save this project and verify it by opening it again.

What you will get

A saved dataset file that has passed a reload consistency check.

Showing 20 of 100
页面连接异常，请刷新后重试。 刷新 🗙