31:T88d,drawGraph={ const width = 600; const height = 400; const padding = 40; const graph = d3.select(DOM.svg(width, height)); const xScale = d3.scaleLinear() .domain([0, d3.max(guess, d => d.time)]) .range([padding, width - padding]); const yScale = d3.scaleLinear() .domain([0, d3.max(guess, d => d.sentences)]) .range([height - padding, padding]) const xAxis = d3.axisBottom() .scale(xScale) .ticks(4) .tickFormat((d)=> (d/1000).toFixed(2)+'s'); const yAxis = d3.axisLeft().scale(yScale).tickFormat(d3.format('.1s')); // text label for the y axis graph.append("text") .attr("transform", "rotate(-90)") .attr("y", 0 - 5) .attr("x",0 - (height / 2)) .attr("dy", "1em") .attr('font-family','sans-serif') .style("text-anchor", "middle") .text("sentences"); // text label for the x axis graph.append("text") .attr("transform", "translate(" + (width/2) + " ," + (height ) + ")") .style("text-anchor", "middle") .text("time"); //draw dot graph.append("circle") .datum(result) .attr('fill','steelblue') .attr("cx", d => xScale(d.time)) .attr("cy", d => yScale(d.sentences)) .attr("r", d => 10); //draw axis #1 graph.append("g") .attr("class", "x axis") .attr("transform", `translate(0, ${height - padding})`) .call(xAxis); //draw axis #2 graph.append("g") .attr("class", "y axis") .attr("transform", `translate(${padding}, 0)`) .call(yAxis); //draw prediction line let coreLine = d3.line().x(d => xScale(d.time)).y(d => yScale(d.sentences)) graph.append("path") .datum(guess) .attr("fill","none") .attr("class","line") .attr("stroke","steelblue") .attr("stroke-width", 3) .attr("stroke-dasharray","5 5") .attr("d", coreLine); // label for prediction-line graph.append("text") .attr("transform", "translate(" + (width/2) + " ," + (height/2.5 ) + ")") .style("text-anchor", "middle") .attr("fill", "steelblue") .style("font-size", 12) .text("(4ms/sentence)"); return graph.node(); }32:T6b1,md`### being very clever: if you're passing high-volumes of text through compromise in speed-sensitive applications, there are some things you can do to get compromise working at top-notch. the first optimisation, if you're on the server-side, is to use the latest version of node. Compromise almost runs twice-as-fast on **node v10** as **v4**. Kodus to node team! ### Caching **[.cache()](https://observablehq.com/@spencermountain/compromise-cache)** allows you to pre-compute a document's words and tags, to fail-fast on some match statements. If you are running many matches, especially on a large document, you may benefit from running \`.cache()\` first, but be wary of [uncovered invalidations](https://observablehq.com/@spencermountain/compromise-cache) if you are also transforming your document. #### Sweep **[.sweep()](https://observablehq.com/@spencermountain/compromise-sweep)** allows you to pre-compute a list of matches, so that no pattern will need to be checked twice, on the same document. This can greatly improve the speed of any exhaustive matches, and can be used to tag a document as well. #### Webworkers To improve performance, or move it off the main thread, you may want to [use a web-worker](https://www.html5rocks.com/en/tutorials/workers/basics/). Because you can't pass-in a function to a web-worker, annoyingly, you must load the library each time. you can see a demo of this [here](https://github.com/spencermountain/compromise/tree/master/demos/web-worker). #### Disabling POS-tagging you can turn-off the most laborious part of compromise, part-of-speech tagging, if you don't need word-categorization, by using \`compromise/one\` or calling \`nlp.tokenize()\` instead: `33:[] 2f:{"title":"compromise-performance","cells":[{"id":0,"value":"md`# compromise-performance\n\ncompromise was built with the goal of parsing and transforming text **on keystroke**, so it is relatively quick.\n\nit is also:\n* **synchronous** - no callbacks!\n* **linear** - *O(n)* - *(10x words take 10x time)* \n\ncompromise can parse ~***one sentence per millisecond*** in a modern desktop browser.\n\nyou can time things here, by loading some large texts, and performing benchmark on them:\n`","pinned":false,"hidden":false,"mode":"ojs"},{"id":5,"value":"viewof txt= html`\n`","pinned":false,"hidden":false,"mode":"ojs"},{"id":11,"value":"{\n return html`\n\n \n\t\n\t\n \n \n\t\n\t\n \n \n\t\n\t\n \n \n\t\n\t\n \n \n\t\n \n

time: \n\t\t${result.time}ms\n\t
\n\t\ttagged \n\t\t${result.words} words\n\t	~${(text.length/1000).toFixed(2)}kb
\n\t\t${result.sentences} sentences:\n\t	${parseFloat(result.time/result.sentences).toFixed(1)}ms / sentence
${result.nouns} nouns	${result.adjectives} adjectives
${result.verbs} verbs

\n`\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":89,"value":"$31","pinned":false,"hidden":false,"mode":"ojs"},{"id":276,"value":"$32","pinned":false,"hidden":false,"mode":"ojs"},{"id":287,"value":"nlp.tokenize('my dog\\'s name is Levi').terms().out('array')","pinned":true,"hidden":false,"mode":"ojs"},{"id":310,"value":"\nmd`### More!\nThere is **a lot** that can be done in terms of caching and performance. Please [get involved](https://github.com/spencermountain/compromise/wiki/Contributing)`","pinned":false,"hidden":false,"mode":"ojs"},{"id":282,"value":"html`

--libraries--`","pinned":false,"hidden":false,"mode":"ojs"},{"id":211,"value":"guess={ return [\n {sentences:1, time:4},\n\n {sentences:500, time:2000},\n ]\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":63,"value":"result=parse(text)","pinned":false,"hidden":false,"mode":"ojs"},{"id":14,"value":"parse=function(str){\n let start=Date.now()\n let doc=nlp(str)\n let end=Date.now()\n return {\n time:end - start,\n words:doc.terms().length,\n sentences:doc.all().length,\n nouns:doc.nouns().length,\n verbs:doc.verbs().length,\n adjectives:doc.match('#Adjective').length,\n }\n return start\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":15,"value":"nlp = require('compromise')","pinned":false,"hidden":false,"mode":"ojs"},{"id":377,"value":"nlp.version","pinned":false,"hidden":false,"mode":"ojs"},{"id":53,"value":"d3 = require(\"https://d3js.org/d3.v5.min.js\")","pinned":false,"hidden":false,"mode":"ojs"},{"id":2,"value":"text=(await fetch(`https://unpkg.com/nlp-corpus@3.0.0/${txt}.txt`)).text()","pinned":false,"hidden":false,"mode":"ojs"},{"id":28,"value":"css=html``","pinned":false,"hidden":false,"mode":"ojs"}],"files":"$W33","stdlib":"1","titling":"auto"}