The (in Jupyter) provided code for “Topic Modeling” had to be run for the full dispatch. In order to extract the full dispatch in the format necessary for using Jupyter, I wrote the following code:

import re
import os

source_location = "./practice3/"   # the folder in which the "dispatch" is
target_location = "./TSV_files/" # the folder in which the new files will be saved 

lof=os.listdir(source_location) # getting all files from a folder
list_items= [] # to append list of items at the end, now out of loop 

up_header = "id\tdate\ttype\theader\ttext\n"


for f in lof: # looping through all the files

    new_file=f + "_modified.xml" # need to add the extension to work with it
    
    with open(source_location + f, "r", encoding="utf8") as f1:
        text0 = f1.read() # reads the file in
        
        ## Extracting the date of the Issue = date of the articles

        date_section = re.search(r"<titlePage>([\n-^\w—]*)</titlePage>", text0).group(1) #finds the section where the date is in
        issue_date=re.search(r"<date value=\"([\d-]*)\" authname",date_section).group(1) #extracts the date of the issue


        ## Getting the articles

        results = re.split(r"<div3 ", text0) # split the text into the div3 parts
        count= 0 # to be able to number the articles later

        for r in results[1:]: #first one is a "library"
            count = count +1
            r ="<div3 " + r # reattach so that it can be canceled out later on
    
            ## type
    
            try:
                item_type=re.search(r"type=\"(\w*)\" ",r).group(1) #check if it has a type, save it
            except:
                item_type="noItemType"
        
            ## header of article
    
            try: 
                header_section= re.search(r"<head>(.*)</head>",r).group(1) #check if it has a header, save it
            except: 
                header_section = "NoHeader"
        
            ## cleaning 
            
            if item_type == "article": 
                
                clean = re.sub("<[^<]+>", "", r) #get rid of all < >
                clean = re.sub(" +\n|\n +","\n",clean) #get rid of all line breaks with spacing
                clean = clean.strip() #get rid of all spacing in beginning and end
                clean = re.sub("\n+","",clean) #get rid of all additional line breaks
                clean = re.sub("\t+","",clean) #get rid of all additional line breaks
                text = clean + "\n"
                header = re.sub("<[^<]+>", "", header_section) #cleaning header of < > 
                ID = issue_date + "_" + item_type + "_" + str(count)
    
            ## collecting all information
    
                if len(re.sub("\W","", clean)) != 0: #noticed that some don't have text but still something -> this way don't include them
                    record= ID + "\t" + issue_date +"\t"+ item_type + "\t" + header + "\t" + text #create record information
                
                    list_items.append(record)    #append to list of all items in this issue

        ## saving files

new_issue ="".join(list_items) #combines the parts

new_file="dispatch_full.tsv" # need to add the extension to work with it
with open(target_location+new_file, "w", encoding ="utf8") as f2: #open in target folder
    f2.write(up_header + new_issue) #write in one file all the information

However, as my computer was unable to perform the analysis with the full dispatch on Jupyter without terminating the program due to too much memory demand/ too long timing - I split the data into separate files for each year, using the following python code:

import re
import os

source_location = "./practice3/"   # the folder in which the "dispatch" is
target_location = "./TSV_files/" # the folder in which the new files will be saved 

lof=os.listdir(source_location) # getting all files from a folder

def tgn(year):
    list_items= []
    up_header = "id\tdate\ttype\theader\ttext\n"
    
    for f in lof: # looping through all the files

        new_file=f + "_modified.xml" # need to add the extension to work with it
    
        with open(source_location + f, "r", encoding="utf8") as f1:
            text0 = f1.read() # reads the file in
        
        ## Extracting the date of the Issue = date of the articles

            date_section = re.search(r"<titlePage>([\n-^\w—]*)</titlePage>", text0).group(1) #finds the section where the date is in
            issue_date=re.search(r"<date value=\"([\d-]*)\" authname",date_section).group(1) #extracts the date of the issue
            issue_year=re.search(r"(\d\d\d\d)",issue_date).group(1)

        ## Getting the articles
            
        results = re.split(r"<div3 ", text0) # split the text into the div3 parts
        count= 0 # to be able to number the articles later

        for r in results[1:]: #first one is a "library"
            count = count +1
            r ="<div3 " + r # reattach so that it can be canceled out later on
    
            ## type
    
            try:
                item_type=re.search(r"type=\"(\w*)\" ",r).group(1) #check if it has a type, save it
            except:
                item_type="noItemType"
        
            ## header of article
    
            try: 
                header_section= re.search(r"<head>(.*)</head>",r).group(1) #check if it has a header, save it
            except: 
                header_section = "NoHeader"
            
            if item_type == "article": 
                if int(issue_year) == int(year):
               
            ## cleaning 
    
                    clean = re.sub("<[^<]+>", "", r) #get rid of all < >
                    clean = re.sub(" +\n|\n +","\n",clean) #get rid of all line breaks with spacing
                    clean = clean.strip() #get rid of all spacing in beginning and end
                    clean = re.sub("\n+","",clean) #get rid of all additional line breaks
                    clean = re.sub("\t+","",clean) #get rid of all additional line breaks
                    text = clean + "\n"
                    header = re.sub("<[^<]+>", "", header_section) #cleaning header of < > 
                    ID = issue_date + "_" + item_type + "_" + str(count)
    
            ## collecting all information
                    if len(re.sub("\W","", clean)) != 0: #noticed that some don't have text but still something -> this way don't include them
                        record= ID + "\t" + issue_date +"\t"+ item_type + "\t" + header + "\t" + text #create record information
                        list_items.append(record)    #append to list of all items in this issue

        ## saving files
                
    new_issue ="".join(list_items) #combines the parts

                
    new_file=str(year) + "_dispatch.tsv" # need to add the extension to work with it
    with open(target_location+new_file, "w", encoding ="utf8") as f2: #open in target folder
        f2.write(up_header + new_issue) #write in one file all the information
        
print ("1861")
tgn(1861)
print ("1862")
tgn(1862)
print ("1863")
tgn(1863)
print ("1864")
tgn(1864)
print ("1865")
tgn(1865)

The observations for different years was compared using once 30 topics, and once 40 topics. In the following, only two years of these are being presented:

The year 1862:

14633 articles

number of words (already “listed” ones): 64112

number of low frequency words (0 or 1): 29876

30 topics:

It took 29 min. 2 sec. to create the model with 30 topics

5 examples for keywords in the topics from the model (detailed) are:

(5,
  '0.017*"war" + 0.012*"service" + 0.010*"military" + 0.009*"department" + '
  '0.008*"corps" + 0.007*"officers" + 0.006*"regiments" + 0.006*"duty" + '
  '0.006*"states" + 0.005*"recruits"')

(29,
  '0.024*"salt" + 0.016*"pound" + 0.013*"articles" + 0.011*"coffee" + '
  '0.011*"sugar" + 0.009*"supply" + 0.009*"price" + 0.008*"gas" + '
  '0.008*"shoes" + 0.008*"whiskey"')

(18,
  '0.080*"cotton" + 0.020*"corn" + 0.018*"tobacco" + 0.017*"crop" + '
  '0.013*"thousand" + 0.013*"wheat" + 0.010*"grain" + 0.010*"bales" + '
  '0.009*"hundred" + 0.008*"flour"')

(25,
  '0.022*"court" + 0.012*"yesterday" + 0.012*"john" + 0.010*"charged" + '
  '0.009*"arrested" + 0.008*"case" + 0.008*"stealing" + 0.007*"mayor" + '
  '0.007*"street" + 0.007*"th"')

(11,
  '0.042*"reward" + 0.020*"feet" + 0.016*"richmond" + 0.015*"negro" + '
  '0.015*"black" + 0.014*"dollars" + 0.014*"boy" + 0.014*"inches" + '
  '0.013*"delivery" + 0.013*"cents"')

All 30 topics can be found below:

{'t00': 'house, mr, night, death, mrs, body, went, room, lady, wife',
 't01': 'general, provost, marshal, order, gen, major, richmond, orders, '
        'martial, military',
 't02': 'government, england, french, mr, london, american, british, th, '
        'states, france',
 't03': 'notes, bank, cent, money, treasury, confederate, paper, bonds, banks, '
        'currency',
 't04': 'iron, navy, guns, virginia, flag, officers, ship, vessels, captain, '
        'naval',
 't05': 'war, service, military, department, corps, officers, regiments, duty, '
        'states, recruits',
 't06': 'river, fort, city, federal, island, th, enemy, miles, morning, point',
 't07': 'states, government, war, state, union, united, president, congress, '
        'country, constitution',
 't08': 'shot, yankee, escape, soldiers, prison, yankees, ball, cut, '
        'execution, young',
 't09': 'fire, prince, oclock, yesterday, royal, funeral, queen, cemetery, '
        'square, alarm',
 't10': 'soldiers, mr, church, city, ladies, rev, richmond, death, life, young',
 't11': 'reward, feet, richmond, negro, black, dollars, boy, inches, delivery, '
        'cents',
 't12': 'south, war, southern, north, states, world, england, government, '
        'northern, country',
 't13': 'enemy, fire, wounded, fight, battery, field, battle, yankees, road, '
        'artillery',
 't14': 'enemy, gen, th, col, wounded, general, battle, killed, regiment, '
        'command',
 't15': 'company, regiment, captain, service, capt, companies, lieutenant, '
        'volunteers, artillery, officers',
 't16': 'mr, bill, state, virginia, governor, senate, house, general, '
        'committee, county',
 't17': 'negroes, homes, god, blood, land, life, hiring, death, heart, love',
 't18': 'cotton, corn, tobacco, crop, thousand, wheat, grain, bales, hundred, '
        'flour',
 't19': 'mr, bill, committee, house, resolution, states, senate, confederate, '
        'congress, military',
 't20': 'army, war, enemy, country, battle, south, southern, field, military, '
        'arms',
 't21': 'tennessee, nashville, dispatch, kentucky, gen, th, news, fort, '
        'federal, northern',
 't22': 'th, co, wounded, company, john, lieut, killed, capt, st, wm',
 't23': 'amp, street, th, mrs, hospital, hire, house, corner, streets, wanted',
 't24': 'army, gen, general, rebel, rebels, mcclellan, richmond, york, '
        'washington, troops',
 't25': 'court, yesterday, john, charged, arrested, case, stealing, mayor, '
        'street, th',
 't26': 'negro, feet, county, inches, city, jail, committed, runaway, aged, '
        'property',
 't27': 'city, orleans, citizens, prisoners, general, gen, butler, government, '
        'order, yankee',
 't28': 'yesterday, prisoners, railroad, train, yankee, county, yankees, '
        'richmond, captured, city',
 't29': 'salt, pound, articles, coffee, sugar, supply, price, gas, shoes, '
        'whiskey'}

When comparing the results with the ones from “mining the dispatch” similarities can be seen. Topic “t11” seems to be equivalent to the topic “fugitive slave ads” and topic “t12” can be considered as “secession”. “t29” might be a topic related to ads but could also be about inflation.

In the following, some exemplatory graphics from the topics are being shown:

For a larger (more frequent) topic:

1862larger30

For a topic of medium (mid frequent) size:

1862medium30

The topic 16 has Virginia, a governor and the senate included and seems to be focused on bills passed there. This topic was very popular in the first half of 1862, however, starting with June there is a sharp drop.

1862t16_30

The topic 25 seems to be including criminal cases held in court the day before, including arrestments and charges. This topic is relevant throughout the whole year of 1862.

1862t25_30

40 topics:

It took 34 min. 29 sec. to create the model with 40 topics

5 examples for keywords in the topics from the model (detailed) are:

(15,
  '0.012*"guns" + 0.009*"fort" + 0.006*"camp" + 0.006*"houston" + 0.006*"duel" '
  '+ 0.006*"manassas" + 0.005*"shouldered" + 0.005*"crenshaw" + 0.004*"sir" + '
  '0.004*"nog"'),

 (11,
  '0.008*"tom" + 0.007*"blind" + 0.004*"hotel" + 0.004*"boarding" + '
  '0.004*"concerts" + 0.003*"soldiers" + 0.003*"house" + 0.003*"mr" + '
  '0.003*"church" + 0.003*"timely"'),

 (0,
  '0.019*"bishop" + 0.012*"bells" + 0.007*"enemy" + 0.007*"cannon" + '
  '0.006*"picquet" + 0.006*"land" + 0.005*"acres" + 0.005*"church" + '
  '0.005*"officer" + 0.004*"sentries"'),

 (39,
  '0.015*"captain" + 0.008*"semmes" + 0.006*"horses" + 0.006*"sumter" + '
  '0.006*"rifle" + 0.005*"vance" + 0.005*"weapon" + 0.005*"used" + '
  '0.004*"capt" + 0.004*"yankee"'),

 (17,
  '0.030*"millions" + 0.027*"salt" + 0.021*"hundred" + 0.017*"dollars" + '
  '0.014*"debt" + 0.010*"tax" + 0.009*"president" + 0.007*"taxation" + '
  '0.007*"property" + 0.006*"pay"')

All 40 topics can be found below:

{'t00': 'bishop, bells, enemy, cannon, picquet, land, acres, church, officer, '
        'sentries',
 't01': 'negroes, servants, hiring, sale, negro, office, servant, estate, '
        'hire, house',
 't02': 'prisoners, yankee, yankees, yesterday, captured, city, richmond, '
        'soldiers, prison, county',
 't03': 'railroad, road, general, company, colonel, miles, bridge, canal, '
        'lieutenant, kanawha',
 't04': 'castle, provost, martial, arrested, marshal, liquor, yesterday, '
        'thunder, godwin, guard',
 't05': 'gen, general, enemy, army, brigade, cavalry, major, command, th, '
        'battle',
 't06': 'wounded, killed, company, privates, slightly, capt, lieut, arm, '
        'severely, private',
 't07': 'enemy, fire, battle, battery, col, guns, regiment, th, wounded, '
        'artillery',
 't08': 'mr, bill, committee, house, senate, resolution, motion, confederate, '
        'military, states',
 't09': 'house, night, street, mr, shot, yesterday, guard, oclock, went, '
        'provost',
 't10': 'army, general, mcclellan, gen, richmond, rebels, rebel, washington, '
        'york, battle',
 't11': 'tom, blind, hotel, boarding, concerts, soldiers, house, mr, church, '
        'timely',
 't12': 'black, page, richmond, feet, dispatch, reward, color, volume, bought, '
        'image',
 't13': 'fort, city, steamer, th, federal, orleans, norfolk, flag, york, '
        'island',
 't14': 'god, confederate, president, states, enemies, thy, davis, prayer, '
        'hand, almighty',
 't15': 'guns, fort, camp, houston, duel, manassas, shouldered, crenshaw, sir, '
        'nog',
 't16': 'street, amp, streets, hire, corner, wanted, th, franklin, hospital, '
        'office',
 't17': 'millions, salt, hundred, dollars, debt, tax, president, taxation, '
        'property, pay',
 't18': 'carolina, north, county, died, state, va, james, local, death, south',
 't19': 'soldiers, ladies, wounded, mrs, sick, hospital, hospitals, amp, '
        'richmond, articles',
 't20': 'england, american, london, france, french, government, british, '
        'english, europe, war',
 't21': 'cents, bank, sales, cotton, market, stock, notes, cent, bonds, prices',
 't22': 'river, canal, james, mr, gunboats, point, petersburg, county, george, '
        'berkeley',
 't23': 'th, co, regiment, st, va, wm, john, lieut, company, ga',
 't24': 'states, government, united, war, state, law, confederate, president, '
        'power, congress',
 't25': 'hall, free, slave, night, metropolitan, varieties, benefit, war, '
        'richmond, illustrations',
 't26': 'enemy, gen, tennessee, force, federal, miles, river, nashville, '
        'cavalry, army',
 't27': 'mr, church, rev, mrs, dr, city, house, virginia, lady, family',
 't28': 'paper, dispatch, virginia, life, editors, citizens, city, lecture, '
        'dr, eastern',
 't29': 'union, mr, state, war, letter, general, york, president, gen, party',
 't30': 'city, meeting, council, death, body, yesterday, night, oclock, '
        'resolved, evening',
 't31': 'committed, norfolk, city, fisher, pairs, sergeant, corporation, '
        'september, james, clothing',
 't32': 'court, john, charged, yesterday, case, stealing, mayor, slave, '
        'arrested, wm',
 't33': 'mr, horse, joe, mrs, genuine, county, city, engraving, election, hill',
 't34': 'reward, feet, negro, inches, th, boy, jail, richmond, black, county',
 't35': 'young, life, lady, thought, tell, heart, poor, am, told, wife',
 't36': 'war, south, southern, country, north, government, enemy, world, '
        'yankee, northern',
 't37': 'iron, navy, river, water, virginia, fire, vessels, work, merrimac, '
        'naval',
 't38': 'service, officers, general, military, company, governor, state, war, '
        'virginia, duty',
 't39': 'captain, semmes, horses, sumter, rifle, vance, weapon, used, capt, '
        'yankee'}

Here, “t38” clearly focuses on the war and the military just like some of the topics of “mining the dispatch”. “t32” seems to be about a criminal charge against a person, a topic that does not seem to be included explicitly in “mining the dispatch”.

In the following, some exemplatory graphics from the topics are being shown:

A larger (more frequent) topic: 1862larger40

A topic of smaller (less frequent) size:

1862smaller40

Topic 3 is focused on railroads and maybe the transportation of troops which seemed to be an important topic in April 1862. Afterwards, it started slowly declining but had another boost towards the end of the year.

1862t16_40

Topic 35 sounds like the topic can be assigned to poetry as some of the keywords are “young, life, lady, heart, poor”. This topic was relevant throughout 1862 with a drop in the month of September.

1862t25_40

The year 1864:

10572 articles

number of words (already “listed” ones): 52344

number of low frequency words (0 or 1): 23270

30 topics:

It took 30 min. 26 sec. to create the model with 30 topics

5 examples for keywords in the topics from the model (detailed) are:

(8,
  '0.046*"priv" + 0.019*"market" + 0.016*"counties" + 0.012*"governor" + '
  '0.009*"butter" + 0.009*"smith" + 0.009*"beef" + 0.009*"bread" + '
  '0.008*"commonwealth" + 0.007*"vacancy"'),

 (17,
  '0.022*"tax" + 0.013*"mr" + 0.010*"collector" + 0.010*"issue" + '
  '0.009*"state" + 0.008*"persons" + 0.008*"paid" + 0.008*"lost" + '
  '0.007*"money" + 0.007*"passport"'),

 (6,
  '0.019*"states" + 0.015*"confederate" + 0.011*"god" + 0.011*"prayer" + '
  '0.010*"government" + 0.009*"almighty" + 0.009*"president" + '
  '0.008*"independence" + 0.008*"public" + 0.008*"worship"'),

 (9,
  '0.068*"st" + 0.035*"january" + 0.027*"profits" + 0.025*"additional" + '
  '0.019*"december" + 0.018*"selling" + 0.018*"buying" + 0.017*"taxes" + '
  '0.016*"cent" + 0.014*"collected"'),

 (11,
  '0.022*"hire" + 0.016*"franklin" + 0.014*"girls" + 0.013*"jpriv" + '
  '0.011*"servants" + 0.011*"women" + 0.011*"wpriv" + 0.011*"encumbrance" + '
  '0.010*"ja" + 0.010*"apriv"')

All 30 topics can be found below:

't00': 'fort, charleston, vessel, fire, water, boat, iron, vessels, guns, '
        'steamer',
 't01': 'reward, dollars, feet, county, th, black, richmond, inches, boy, '
        'negro',
 't02': 'war, states, lincoln, president, union, peace, state, mr, government, '
        'general',
 't03': 'government, states, mr, united, th, york, war, england, confederate, '
        'london',
 't04': 'bonds, treasury, states, confederate, exchange, cent, department, '
        'treasurer, richmond, america',
 't05': 'navy, notice, th, office, special, oclock, richmond, john, va, '
        'service',
 't06': 'states, confederate, god, prayer, government, almighty, president, '
        'independence, public, worship',
 't07': 'general, maryland, virginia, army, confederate, state, soldiers, war, '
        'service, duty',
 't08': 'priv, market, counties, governor, butter, smith, beef, bread, '
        'commonwealth, vacancy',
 't09': 'st, january, profits, additional, december, selling, buying, taxes, '
        'cent, collected',
 't10': 'dispatch, cents, prices, price, richmond, pounds, pound, bushel, '
        'morning, horses',
 't11': 'hire, franklin, girls, jpriv, servants, women, wpriv, encumbrance, '
        'ja, apriv',
 't12': 'court, charged, mrs, slave, stealing, john, case, yesterday, james, '
        'miss',
 't13': 'gen, th, enemy, wounded, brigade, cavalry, killed, col, artillery, '
        'battle',
 't14': 'sale, th, co, cotton, tobacco, street, auction, sugar, fine, '
        'furniture',
 't15': 'prisoners, yankee, confederate, city, yesterday, castle, thunder, '
        'arrested, prison, committed',
 't16': 'reward, dollars, richmond, feet, hundred, negro, inches, th, '
        'delivery, black',
 't17': 'tax, mr, collector, issue, state, persons, paid, lost, money, '
        'passport',
 't18': 'notes, treasury, certificates, authorized, secretary, payable, '
        'government, states, commonwealth, payment',
 't19': 'city, co, th, order, mr, oclock, members, hall, richmond, st',
 't20': 'mr, shot, killed, night, body, train, city, house, county, negro',
 't21': 'th, oclock, residence, friends, funeral, attend, family, died, aged, '
        'church',
 't22': 'officers, general, richmond, company, orders, virginia, forces, '
        'order, command, reserve',
 't23': 'service, officer, persons, enrolling, th, officers, general, '
        'exemption, bureau, district',
 't24': 'mr, bill, committee, house, senate, resolution, confederate, '
        'congress, states, referred',
 't25': 'court, th, judge, city, jury, oclock, guilty, wm, trial, term',
 't26': 'war, yankee, country, army, yankees, world, south, take, battle, life',
 't27': 'house, mr, street, night, negro, morning, fire, yesterday, mayor, '
        'room',
 't28': 'hire, wanted, house, sale, cook, street, servant, office, woman, th',
 't29': 'general, enemy, army, hundred, river, force, gen, troops, miles, th'}

The topic “t14” seems to be talking about the trade with sugar and cotton which is involving slaves. This seems not to be mentioned within the “mining the dispatch”. However, again, the topic war reports and military campaigns is in “mining the dispatch” and here within the topic “t00” inlcuded.

In the following, some exemplatory graphics from the topics are being shown:

A larger (more frequent) topic:

1864larger30

A smaller (less frequent) topic:

1864smaller30

The topic 19 is talking about the city of Richmond. Also members and the hall is being mentioned, suggesting that this topic includes important meetings of the Confederate government. This seems to be very important in the month of April with a high peak, then declining and being stable for the rest of the year 1864.

1864t19_30

Topic 20 seems to be focused on incidnts where somebody was shot and killed during the night in a county, yielding the impression that this topic includes murder cases, however, unclear if lynching murders, stopped escape attempts or attacks. This topic seemed to be relevant throughout the year, however, especially at the beginning of 1864.

1864t20_30

40 topics:

It took 33 min. 27 sec. to create the model with 40 topics

5 examples for keywords in the topics from the model (detailed) are:

(22,
  '0.029*"slaves" + 0.025*"property" + 0.015*"owner" + 0.014*"impressment" + '
  '0.012*"officer" + 0.010*"impressed" + 0.010*"according" + 0.009*"section" + '
  '0.009*"war" + 0.009*"value"'),

 (35,
  '0.013*"state" + 0.012*"claims" + 0.012*"service" + 0.012*"irish" + '
  '0.011*"confederate" + 0.010*"ireland" + 0.009*"soldiers" + 0.008*"hill" + '
  '0.008*"states" + 0.006*"government"'),

 (5,
  '0.011*"county" + 0.008*"dress" + 0.008*"belongs" + 0.008*"captured" + '
  '0.007*"horse" + 0.007*"silk" + 0.007*"horses" + 0.006*"white" + '
  '0.005*"dahlgren" + 0.005*"fine"'),

 (32,
  '0.028*"th" + 0.018*"april" + 0.014*"confederate" + 0.012*"taxes" + '
  '0.011*"public" + 0.009*"states" + 0.009*"march" + 0.009*"payment" + '
  '0.009*"persons" + 0.009*"subsistence"'),
  
 (19,
  '0.043*"company" + 0.033*"express" + 0.032*"priv" + 0.028*"southern" + '
  '0.019*"soldiers" + 0.016*"associations" + 0.014*"quartermaster" + '
  '0.014*"richmond" + 0.013*"virginia" + 0.013*"army"'),)

All 40 topics can be found below:

't00': 'army, grant, general, gen, battle, war, lee, country, richmond, ' 'yankee', 
't01': 'war, government, states, mr, peace, united, country, lincoln, state, ' 'president', 
't02': 'soldiers, children, women, citizens, city, poor, families, homes, ' 'home, war', 
't03': 'bonds, treasury, notes, cent, states, confederate, secretary, ' 'certificates, department, treasurer', 
't04': 'church, city, ladies, dr, young, hall, evening, night, rev, oclock', 
't05': 'county, dress, belongs, captured, horse, silk, horses, white, ' 'dahlgren, fine', 
't06': 'fort, vessels, vessel, steamer, charleston, ship, captain, boat, ' 'fleet, island', 
't07': 'mr, bill, committee, house, senate, resolution, confederate, ' 'military, referred, congress', 
't08': 'association, press, telegraph, northern, office, reports, work, ' 'richmond, georgia, clerks', 
't09': 'th, oclock, residence, friends, funeral, attend, died, family, ' 'invited, aged', 
't10': 'dispatch, cents, morning, richmond, subscribers, daily, week, ' 'carriers, charge, news', 
't11': 'court, judge, confederate, states, case, district, taxed, habeas, ' 'corpus, cases', 
't12': 'maryland, general, camp, major, colonel, virginia, johnson, line, ' 'regiment, staunton', 
't13': 'atlanta, railroad, army, tennessee, georgia, sherman, enemy, miles, ' 'general, river', 
't14': 'officers, service, enrolling, persons, officer, general, orders, th, ' 'state, bureau', 
't15': 'enemy, gen, wounded, cavalry, yesterday, killed, line, th, night, ' 'prisoners', 
't16': 'city, farm, market, richmond, office, county, miles, frederick, ' 'company, usual', 
't17': 'prisoners, yankee, prison, yesterday, libby, flag, truce, confined, ' 'officers, deserters', 
't18': 'house, street, night, mr, th, store, room, oclock, lot, streets', 
't19': 'company, express, priv, southern, soldiers, associations, ' 'quartermaster, richmond, virginia, army', 
't20': 'reward, dollars, feet, inches, richmond, black, negro, county, ' 'delivery, hundred', 
't21': 'hire, wanted, house, cook, sale, servant, price, street, ja, prices', 
't22': 'slaves, property, owner, impressment, officer, impressed, according, ' 'section, war, value', 
't23': 'co, th, inst, slave, hat, richmond, gray, tobacco, horse, coat', 
't24': 'th, co, capt, company, st, gen, col, regiment, va, lieut', 
't25': 'carolina, north, virginia, south, georgia, thomas, city, county, ' 'vance, alabama', 
't26': 'fire, mr, hundred, dollars, thousand, destroyed, negroes, corn, work, ' 'burnt', 
't27': 'mrs, miss, john, wm, co, dr, james, mr, rev, smith', 
't28': 'st, january, profits, additional, selling, buying, tax, cent, taxes, ' 'collected', 
't29': 'cotton, sale, sugar, shoes, black, goods, street, dozen, french, ' 'english', 
't30': 'governor, county, commonwealth, virginia, richmond, smith, court, th, ' 'price, state', 
't31': 'charged, yesterday, stealing, court, arrested, slave, mayor, negro, ' 'charge, mr', 
't32': 'th, april, confederate, taxes, public, states, march, payment, ' 'persons, subsistence', 
't33': 'virginia, city, institute, cadets, richmond, seventh, river, cavalry, ' 'hunter, hospital', 
't34': 'exchange, richmond, th, sale, auction, city, co, sell, hill, ' 'confederate', 
't35': 'state, claims, service, irish, confederate, ireland, soldiers, hill, ' 'states, government', 
't36': 'states, confederate, navy, oclock, almighty, america, president, ' 'prayer, public, god', 
't37': 'london, french, th, france, paris, england, europe, king, english, ' 'american', 
't38': 'certificates, government, treasury, taxation, states, payable, ' 'officers, authorized, supplies, required', 
't39': 'general, army, th, rebel, york, rebels, hundred, thousand, ' 'washington, city'} 

Topic “t38” focuses on taxation, certificates and government, making it comparable with the topic “war bonds” from “mining the dispatch”. The same holds for the topic “t15” and “military campaigns”.

In the following, some exemplatory graphics from the topics are being shown:

Larger (more frequent) topic:

1864larger40

Smaller (less frequent) topic:

1864smaller40

Topic 3 is focused on bonds, treasury and the state. This topic is probably related to war bonds, which would also explained the sharp rise of the topic over the last months of 1864, as the end of the Civil War is coming closer and the money demands of the Southern government are increasing.

1864t3_40

Topic 20 seems to be about fugitive slaves. The keywords rewards, dollars, feet, balck, negro and delivery are suggesting this assumption. As the graph shows, this topic seems to be important throughout the years with several higher peaks, maybe when several slaves had escaped. Especially towards the end of the war some slaves escaped up North to take up arms against the Confederacy.

1864t20_40