diff options
author | Maurice Laveaux <m.laveaux@student.tue.nl> | 2014-05-19 11:13:35 +0200 |
---|---|---|
committer | Maurice Laveaux <m.laveaux@student.tue.nl> | 2014-05-19 11:13:35 +0200 |
commit | b1bbd92fdcf9f52d38852b6c5ae802cc401ff712 (patch) | |
tree | 7ece7058a0ca39d4773b18e70437b8978aa3022a /src | |
parent | 058ef3d39029da492477c3c59f1e405193ca3013 (diff) | |
parent | 1f5085cb29caa40c96878dd0fc4194abdb2cac2b (diff) | |
download | Goldfarmer-b1bbd92fdcf9f52d38852b6c5ae802cc401ff712.tar.gz |
Merge branch 'master' of git.lekensteyn.nl:tue/2IOC0-DBL/Goldfarmer
Diffstat (limited to 'src')
-rw-r--r-- | src/database/QueryUtils.java | 6 | ||||
-rw-r--r-- | src/main/Analyzor.java | 509 | ||||
-rw-r--r-- | src/main/FarmShell.java | 4 |
3 files changed, 255 insertions, 264 deletions
diff --git a/src/database/QueryUtils.java b/src/database/QueryUtils.java index 9aab081..2cc6fd6 100644 --- a/src/database/QueryUtils.java +++ b/src/database/QueryUtils.java @@ -99,4 +99,10 @@ public class QueryUtils { } + public static void setInsertBrandParams(NamedPreparedStatement brandStmt, + long id, String brand) throws SQLException { + brandStmt.setLong("tweetid", id); + brandStmt.setString("brand", brand); + // TODO: rating (positive) + } } diff --git a/src/main/Analyzor.java b/src/main/Analyzor.java index 39df43c..aa2d0a4 100644 --- a/src/main/Analyzor.java +++ b/src/main/Analyzor.java @@ -1,264 +1,245 @@ -package main; - -import database.NamedPreparedStatement; -import database.QueryUtils; -import java.io.File; -import java.io.FileNotFoundException; -import java.io.IOException; -import java.io.PrintWriter; -import java.io.UnsupportedEncodingException; -import java.sql.Connection; -import java.sql.PreparedStatement; -import java.sql.ResultSet; -import java.sql.SQLException; -import java.util.HashMap; -import java.util.Scanner; - -/** - * The sentiment analysis class that rates tweets based on a unigram and bigram - * set of weights. - */ -public class Analyzor { - - /** - * The map that matches single words to their weights. - */ - private final HashMap<String, Double> unimap = new HashMap(); - - /** - * The map that matches word pairs to their weights. - */ - private final HashMap<String, Double> bimap = new HashMap(); - - /** - * The results of the query. - */ - private ResultSet data; - - /** - * The existing connection to the database. - */ - private final Connection connection; - - /** - * @param connection A connection to the database. - */ - public Analyzor(Connection connection) { - this.connection = connection; - } - - /** - * Read the unigram and bigram lexica. - * - * @throws FileNotFoundException - */ - public void readLexicon() throws FileNotFoundException { - if (!unimap.isEmpty()) { - // data is already read. - return; - } - System.err.println("Trying to read lexicons..."); - // A unigram is in the format (WS = whitespace): - // word <WS> rating <WS> ??? <WS> ?? - // A bigram has an two WS-separated words instead of one. - try (Scanner uniScanner = new Scanner(new File("unigrams-pmilexicon.txt")); - Scanner biScanner = new Scanner(new File("bigrams-pmilexicon.txt"));) { - //Fill the map of unigrams - while (uniScanner.hasNext()) { - String words = uniScanner.next(); - unimap.put(words.toLowerCase(), uniScanner.nextDouble()); - if (uniScanner.hasNextLine()) { - uniScanner.nextLine(); - } - } - - //fill the map of bigrams - while (biScanner.hasNext()) { - String words = biScanner.next() + " " + biScanner.next(); - bimap.put(words.toLowerCase(), biScanner.nextDouble()); - if (biScanner.hasNextLine()) { - biScanner.nextLine(); - } - } - } - System.err.println("Lexicons are read."); - } - - /** - * Executes a query that the analyzer can analyze. - * - * @param query The query string to execute. - * @throws SQLException When database connection isn't available. - */ - public void query(String query) throws SQLException { - PreparedStatement statement; - //make a connection to the database and execute the query - statement = connection.prepareStatement(query); - data = statement.executeQuery(); - } - - /** - * Run a sentiment analysis and fill the database with the output. - * - * @param query The query to analyze. - * @throws SQLException - * @throws IOException - */ - public void sentimentAnalysis(String query) throws SQLException, IOException { - query(query); - - //read the lexicons - readLexicon(); - - //go to the start of te dataset - if (data == null) { - System.err.println("data is empty, try querying first"); - return; - } - - Double value; - String text; - - //for all tuples - while (data.next()) { - //get the text - text = data.getString("text"); - text = splitPunctToWords(text); - // test is the tweet text you are going to analyze - String[] words = text.split("\\s+"); // text splitted into separate words - double positiverate = 0; // positive rating - - // Rate the text with unigrams - for (String word : words) { - value = unimap.get(word); - if (value != null) { - positiverate += unimap.get(word); - } - } - // Rate the text with bigrams - for (int i = 0; i < words.length - 1; i++) { - String pair = words[i] + " " + words[i + 1]; - value = bimap.get(pair); - if (value != null) { - positiverate += bimap.get(pair); - } - } - //insert the rating into the database - NamedPreparedStatement m_insertRating; - m_insertRating = new NamedPreparedStatement(connection, QueryUtils.insertRating); - QueryUtils.setInsertParams(m_insertRating, data.getLong("tweetid"), data.getString("brand"), (int) (positiverate * 10)); - m_insertRating.executeUpdate(); - //don't print the rate - //System.out.println(text + ": " + (int) (positiverate * 10)); - } - } - - /** - * Make a word cloud of the results of a query. - * - * @param query - * @throws SQLException - * @throws FileNotFoundException - * @throws UnsupportedEncodingException - */ - public void makeWordCloud(String query) throws SQLException, FileNotFoundException, UnsupportedEncodingException { - - query(query); - //go to the start of the ResultSet data - if (data == null) { - System.err.println("data is empty, try querying first"); - return; - } - - String text; - String[] words; - Integer value; - String tweetid; - - PrintWriter writer = new PrintWriter("wordcloud.csv", "UTF-8"); - //print the first row - writer.println("tweetid, word"); - - while (data.next()) { - //get the text - text = data.getString("text"); - //remove punctuation, convert to lowercase and split on words - text = removePunct(text); - text = text.toLowerCase(); - words = text.split("\\s+"); - //we use the tweetid as case id - tweetid = Long.toString(data.getLong("tweetid")); - - for (String word : words) { - writer.println(tweetid + ", " + word); - } - } - //print it in a csv file to put in disco - - //print the first row - - //print the values - writer.close(); - } - - /** - * Makes a comma seperated values file from the query and writes it to output.csv - * - * @param query The string of the query to execute. - * @throws SQLException - * @throws FileNotFoundException - * @throws UnsupportedEncodingException - */ - public void disco(String query) throws SQLException, FileNotFoundException, UnsupportedEncodingException { - //do the query - query(query); - PrintWriter writer = new PrintWriter("output.csv", "UTF-8"); - //print the first row - for (int i = 1; i < data.getMetaData().getColumnCount(); i++) { - writer.print(data.getMetaData().getColumnLabel(i) + ", "); - } - writer.println(data.getMetaData().getColumnLabel(data.getMetaData().getColumnCount())); - //print the values - while (data.next()) { - for (int i = 1; i < data.getMetaData().getColumnCount(); i++) { - if (data.getObject(i) == null) { - writer.print(", "); - } else { - writer.print(data.getObject(i).toString().replaceAll("[,\n]", " ") + ", "); - } - } - if(data.getObject(data.getMetaData().getColumnCount())==null){ - writer.println("0"); - } else { - writer.println(data.getObject(data.getMetaData().getColumnCount()).toString().replace(",", " ")); - } - } - writer.close(); - } - - /** - * Replaces punctuation so it can be split correctly. - * - * @param text The input text - * @return The text with punctuation replaced. - */ - private String splitPunctToWords(String text) { - text = text.replaceAll("https?://\\S*", ""); - text = text.replaceAll("[!?):;\"']", " $0"); - text = text.replaceAll("[.,-](\\s|$)", " $0"); - text = text.replaceAll("\\s[(\"']", "$0 "); - return text; - } - - /** - * Removes punctuation and urls. - * - * @param text The input text - * @return The text with punctuation removed. - */ - private String removePunct(String text) { - text = text.replaceAll("https?://\\S*", ""); - text = text.replaceAll("[.,!?()-:;\"']", " "); - return text; - } -} +package main;
+
+import analysis.BrandChecker;
+import database.NamedPreparedStatement;
+import database.QueryUtils;
+import java.io.File;
+import java.io.FileNotFoundException;
+import java.io.IOException;
+import java.io.PrintWriter;
+import java.io.UnsupportedEncodingException;
+import java.sql.Connection;
+import java.sql.PreparedStatement;
+import java.sql.ResultSet;
+import java.sql.SQLException;
+import java.util.HashMap;
+import java.util.Map.Entry;
+import java.util.Scanner;
+
+/**
+ * The sentiment analysis class that rates tweets based on a unigram and bigram
+ * set of weights.
+ */
+public class Analyzor {
+
+ /**
+ * The map that matches single words to their weights.
+ */
+ private final HashMap<String, Double> unimap = new HashMap();
+
+ /**
+ * The map that matches word pairs to their weights.
+ */
+ private final HashMap<String, Double> bimap = new HashMap();
+
+ private ResultSet data;
+ private final Connection connection;
+
+ Analyzor(Connection connection) {
+ this.connection = connection;
+ }
+
+ //reads the lexicons
+ void readLexicon() throws FileNotFoundException {
+ if (!unimap.isEmpty()) {
+ // data is already read.
+ return;
+ }
+ System.err.println("Trying to read lexicons...");
+ // A unigram is in the format (WS = whitespace):
+ // word <WS> rating <WS> ??? <WS> ??
+ // A bigram has an two WS-separated words instead of one.
+ try (Scanner uniScanner = new Scanner(new File("unigrams-pmilexicon.txt"));
+ Scanner biScanner = new Scanner(new File("bigrams-pmilexicon.txt"));) {
+ //Fill the map of unigrams
+ while (uniScanner.hasNext()) {
+ String words = uniScanner.next();
+ unimap.put(words.toLowerCase(), uniScanner.nextDouble());
+ if (uniScanner.hasNextLine()) {
+ uniScanner.nextLine();
+ }
+ }
+
+ //fill the map of bigrams
+ while (biScanner.hasNext()) {
+ String words = biScanner.next() + " " + biScanner.next();
+ bimap.put(words.toLowerCase(), biScanner.nextDouble());
+ if (biScanner.hasNextLine()) {
+ biScanner.nextLine();
+ }
+ }
+ }
+ System.err.println("Lexicons are read.");
+ }
+
+ /**
+ * Executes a query that the analyzer can analyze.
+ *
+ * @param query The query string to execute.
+ * @throws SQLException When database connection isn't available.
+ */
+ public void query(String query) throws SQLException {
+ PreparedStatement statement;
+ //make a connection to the database and execute the query
+ statement = connection.prepareStatement(query);
+ data = statement.executeQuery();
+ }
+
+ /**
+ * Run a sentiment analysis and fill the database with the output.
+ *
+ * @throws SQLException
+ * @throws IOException
+ */
+ public void sentimentAnalysis(String query) throws SQLException, IOException {
+ query(query);
+
+ //read the lexicons
+ readLexicon();
+
+ //go to the start of te dataset
+ if (data == null) {
+ System.err.println("data is empty, try querying first");
+ return;
+ }
+
+ Double value;
+ String text;
+
+ //for all tuples
+ while (data.next()) {
+ //get the text
+ text = data.getString("text");
+ text = splitPunctToWords(text);
+ // test is the tweet text you are going to analyze
+ String[] words = text.split("\\s+"); // text splitted into separate words
+ double positiverate = 0; // positive rating
+
+ // Rate the text with unigrams
+ for (String word : words) {
+ value = unimap.get(word);
+ if (value != null) {
+ positiverate += unimap.get(word);
+ }
+ }
+ // Rate the text with bigrams
+ for (int i = 0; i < words.length - 1; i++) {
+ String pair = words[i] + " " + words[i + 1];
+ value = bimap.get(pair);
+ if (value != null) {
+ positiverate += bimap.get(pair);
+ }
+ }
+ //insert the rating into the database
+ NamedPreparedStatement m_insertRating;
+ m_insertRating = new NamedPreparedStatement(connection, QueryUtils.insertRating);
+ QueryUtils.setInsertParams(m_insertRating, data.getLong("tweetid"), data.getString("brand"), (int) (positiverate * 10));
+ m_insertRating.executeUpdate();
+ //don't print the rate
+ //System.out.println(text + ": " + (int) (positiverate * 10));
+ }
+ }
+
+ //makes a wordcloud of the tweets in the ResultSet data
+ void makeWordCloud(String query) throws SQLException, FileNotFoundException, UnsupportedEncodingException {
+
+ query(query);
+ //go to the start of the ResultSet data
+ if (data == null) {
+ System.err.println("data is empty, try querying first");
+ return;
+ }
+
+ String text;
+ String[] words;
+ Integer value;
+ String tweetid;
+
+ PrintWriter writer = new PrintWriter("wordcloud.csv", "UTF-8");
+ //print the first row
+ writer.println("tweetid, word");
+
+ while (data.next()) {
+ //get the text
+ text = data.getString("text");
+ //remove punctuation, convert to lowercase and split on words
+ text = removePunct(text);
+ text = text.toLowerCase();
+ words = text.split("\\s+");
+ //we use the tweetid as case id
+ tweetid = Long.toString(data.getLong("tweetid"));
+
+ for (String word : words) {
+ writer.println(tweetid + ", " + word);
+ }
+ }
+ //print it in a csv file to put in disco
+
+ //print the first row
+
+ //print the values
+ writer.close();
+ }
+
+ //generate csv for disco from the query
+ public void disco(String query) throws SQLException, FileNotFoundException, UnsupportedEncodingException {
+ //do the query
+ query(query);
+ PrintWriter writer = new PrintWriter("output.csv", "UTF-8");
+ //print the first row
+ for (int i = 1; i < data.getMetaData().getColumnCount(); i++) {
+ writer.print(data.getMetaData().getColumnLabel(i) + ", ");
+ }
+ writer.println(data.getMetaData().getColumnLabel(data.getMetaData().getColumnCount()));
+ //print the values
+ while (data.next()) {
+ for (int i = 1; i < data.getMetaData().getColumnCount(); i++) {
+ if (data.getObject(i) == null) {
+ writer.print(", ");
+ } else {
+ writer.print(data.getObject(i).toString().replaceAll("[,\n]", " ") + ", ");
+ }
+ }
+ if(data.getObject(data.getMetaData().getColumnCount())==null){
+ writer.println("0");
+ } else {
+ writer.println(data.getObject(data.getMetaData().getColumnCount()).toString().replace(",", " "));
+ }
+ }
+ writer.close();
+ }
+
+ public void getBrands() throws SQLException{
+ PreparedStatement statement;
+ //make a connection to the database and execute the query
+ statement = connection.prepareStatement("delete from mentionsbrand");
+ statement.executeUpdate();
+ BrandChecker checker = new BrandChecker("brandrules.txt");
+ query("select * from tweet");
+ NamedPreparedStatement m_insertBrand = new NamedPreparedStatement(connection, QueryUtils.insertBrand);
+ while(data.next()){
+ for(String brand:checker.getBrands(data.getString("text"))){
+ QueryUtils.setInsertBrandParams(m_insertBrand, data.getLong("tweetid"), brand);
+ m_insertBrand.executeUpdate();
+ }
+ }
+ }
+
+ //replaces punctuation so it will be splitted
+ //also removes urls
+ private String splitPunctToWords(String text) {
+ text = text.replaceAll("https?://\\S*", "");
+ text = text.replaceAll("[!?):;\"']", " $0");
+ text = text.replaceAll("[.,-](\\s|$)", " $0");
+ text = text.replaceAll("\\s[(\"']", "$0 ");
+ return text;
+ }
+
+ //removes punctuation
+ //also removes urls
+ private String removePunct(String text) {
+ text = text.replaceAll("https?://\\S*", "");
+ text = text.replaceAll("[.,!?()-:;\"']", " ");
+ return text;
+ }
+}
diff --git a/src/main/FarmShell.java b/src/main/FarmShell.java index 044d3d3..1266fd3 100644 --- a/src/main/FarmShell.java +++ b/src/main/FarmShell.java @@ -133,6 +133,9 @@ public class FarmShell { case disco: getAnalyzor().disco(params[0]); break; + case getBrands: + getAnalyzor().getBrands(); + break; case help: for (String line : HELP) { System.out.println(line); @@ -162,6 +165,7 @@ public class FarmShell { filterbots("marks all users as bot or not", 1), sentiment("analyzes all tweets on positivity (about a brand)", 1), wordcloud("makes a wordcloud of the text of the tweets", 1), + getBrands("fills the database with the brands of a tweet"), disco("makes a outputfile for disco",1), exit("Returns to shell"), help("Get help"); |