Social scientific studies

Several past works have studied the portrayal of professions in entertainment media. Asimow studied the representation of lawyers in 284 films and found most portrayals to be negative [16]. Dimnik et al. examined the representation of accountants in 121 movies and extracted six character stereotypes of the accountant personality [17]. Flores investigated physicians’ image depictions in 131 films and found that they were mostly depicted as greedy and uncaring [18]. Pautz used a sample of 34 films containing more than 200 police (cop) characters and found that most cops were shown as good, hard-working, and competent law-enforcement officers [19]. Kalisch et al. analyzed 670 nurse and 466 physician characters in novels, movies and television series, and concluded that compared to physicians, nurses portrayed in the media were consistently less central to the plot, less intelligent, less rational, and less likely to exercise clinical judgement [29]. Smith et al. investigated gender representation of occupations in films, prime-time programs, and children TV shows, and found that women are grossly underrepresented compared to men in science, technology, engineering and math jobs (STEM) [30]. These works involved extensive human coding and profession-specific analysis which is not reproducible on a large scale. The present study offers a complementary view in the sense it trades off a smaller scale character-centric study for a larger scale lexical analysis, focusing on character utterances instead of personality traits. The remaining sections describe the computational methods used to achieve the same.

Named entity recognition

Named entity recognition (NER) is a classic NLP task of finding entity mentions in text and classifying their type. Traditional NER primarily targets person, organization, and location entity types. Most NER datasets only contain labels for these three types of entities [31]. The OntoNotes 5 dataset increased the number of entity types to 18 by including nationalities, products, events, and numeric values [32]. However, it does not contain any professional titles. Fine-grained NER extends traditional entity recognition by expanding the entity set to include hundreds of named categories. Ling et al. created a benchmark dataset for fine-grained NER that labels 112 different entity types but it only contains a few professional titles [33]. Sekine built an ontology for named entities and defined professional titles as vocational attributes for the person-entity type [34]. Mai et al. used this entity hierarchy to annotate English and Japanese sentences and evaluated different fine-grained NER models [35]. However, they did not include the professional attributes in their labeling set. The Text Analysis Conference Knowledge Base Population (TAC KBP) track of entity discovery and linking introduced job titles as entity types to model the person:title relationship [36]. The Stanford CoreNLP NER model used the KBP 2017 dataset to create regular expression-based rules for finding professional titles in text [37]. However, we observed that it missed many of the professional mentions in media text.

In the absence of labeled data, entity gazetteers are often used to find candidate spans for named entities. Gazetteers are curated lists of entities that improve NER performance when combined with supervised models. Lin et al. used gazetteers to identify text subsequences for the region-based encoder and improved the state-of-the-art NER performance on the ACE2005 benchmark [38]. Liu et al. combined dictionary lookups with semi-Markov CRF (conditional random field) architectures and achieved comparable results with more complex neural models on the CoNLL 2003 and OntoNotes 5 datasets [39]. Several gazetteers of job titles are available (Gate job titles, fluquid). Government and international organizations use some standard gazetteers of job titles to collect occupational data. For example, the International Labor Organization uses the International Standard Classification of Occupations [40], and the US Bureau of Labor Statistics maintains the Standard Occupational Classification (SOC) system [25]. Aside from careful human coding, such gazetteers can be constructed using different automated methods, including those that leverage existing knowledge bases, such as Wikipedia and WordNet [41, 42]. In this work, we use the SOC taxonomy to get the initial list of job titles and professional groups.

WordNet

WordNet is a widely-used lexical resource in English [43]. It groups words into synonym sets called synsets. Synonymy is the semantic relationship between words of similar meaning. Polysemous words have multiple meanings and belong to more than one synset. For example, the word “conductor” is present in three synsets—conductor.n.01 (the person who leads a musical group), conductor.n.02 (a substance that readily conducts electricity and heat) and conductor.n.03 (the person who collects fares on a public conveyance). The name of a synset is composed of three dot-separated literals. The first literal is the lemmatized form of the main word of the synset, the second literal is the part-of-speech, and the third literal is the sense index. Synsets are tagged by semantic classes [44]. Synset A is a hyponym of synset B if it is a more specific form of B. For example, allergist.n.01 (doctor specialized in the treatment of allergies), surgeon.n.01 (doctor specialized in surgery) and veterinarian.n.01 (doctor practicing veterinary medicine) are all hyponyms of the more general synset, doctor.n.01 (a licensed medical practitioner).

WordNet has been used to construct entity gazetteers. Toral et al. leveraged WordNet’s noun hierarchy to build person and location gazetteers [45]. Maginini et al. used WordNet to identify trigger words and gazetteer terms for English NER [46]. Boteanu et al. expanded a shopping taxonomy for efficient product search by matching the product names to WordNet synsets [47]. In this work, we use WordNet synsets to extend the SOC taxonomy and create a searchable dictionary of professions. WordNet synsets are also the target labels for the word sense disambiguation task, an integral part of our search pipeline.

Word sense disambiguation

Word sense disambiguation (WSD) is the task of assigning words in context to their most appropriate sense. Wordnet usually serves as the sense inventory that provides the target senses. The same word can express different meanings depending upon its context. Consider the word “conductor” in the following two sentences—“Conductors communicate with the musicians through hand gestures” and “Metals are good heat conductors”. The former refers to a person directing the music of an orchestra, denoted by the synset conductor.n.01. The latter means a heat-conducting substance, denoted by the synset conductor.n.02 (See Sec WordNet). Many NLP tasks such as machine translation, information retrieval and question answering use WSD in their text-processing pipeline [48–51].

Knowledge-based and supervised approaches have tackled WSD, with the latter usually outperforming the former. Raganato et al. standardized the evaluation framework for WSD and used a combination of SenseEval [52, 53] and SemEval [54–56] datasets to compare the performance of different WSD models [57]. Raganato et al. treated WSD as a sequence labeling task and used an LSTM with attention layers to find the sense of all sentence words jointly [58]. Huang et al. constructed context-gloss pairs and converted WSD into a sentence pair classification task [59]. Kumar et al. produced gloss embeddings using the WordNet graph and combined them with contextual vectors to find the word sense [60]. Bevilacqua et al. extended this method by adding hypernym and hyponym relational knowledge to construct the synset vectors and achieved state-of-the-art WSD performance [61]. In this work, we use WSD to remove non-professional mentions of job titles in media subtitles.

Sentiment analysis

Sentiment analysis or opinion mining is the task of finding the sentiments, opinions, attitudes, appraisals and emotions towards entities or their attributes expressed or implied in text [62]. Sentiment is always targeted at some entity or towards some attribute of the entity. The target entity or attribute can be a person, organization, issue, product, service, topic or event. Such target-oriented opinion mining is called aspect-based sentiment analysis (ABSA).

We use professional mentions as opinion targets for sentiment analysis to find how positively or negatively different professions are talked about in media stories. The task of profession ABSA is to find the sentiment orientation of the opinion expressed towards the person or group of persons referred to by their job title. If the job title does not refer to any profession, or people employed in the profession, the sentiment is deemed neutral. The following example sentences show the sentiment label of the job title word, marked in bold.

1. Harry Floyd was a great actor. (positive)
   Explanation: Actor refers to Harry Floyd who is described as great.
2. But that damn vet kept ordering test after test after test! (negative)
   Explanation: The speaker uses a swear term, damn, to address the veterinarian and criticizes his or her action.
3. Fine, then we get the armor and reverse engineer it. (Neutral)
   Explanation: The word engineer refers to an action, not a profession.
4. You’re going to be a lousy architect. (Neutral)
   Explanation: The person towards which the negative sentiment is expressed, is not yet an architect.

Benchmark ABSA datasets exist for several domains such as question answering forums, customer reviews and tweets [28, 63, 64]. Dong et al. proposed an adaptive recursive neural network for target-dependent twitter sentiment classification, that propagated the sentiments of words to their targets depending upon the context and syntactic relations [28]. Tang et al. used LSTMs to model the left and right context of the target entity for twitter sentiment classification [65]. Memory networks and graph convolutional neural models have also been proposed [66, 67]. Recently, several transformer networks have been introduced for the ABSA task and have achieved state-of-the-art performance [68]. Sun et al. constructed sentences containing the aspect expression and the sentiment orientation, and fine-tuned a pre-trained BERT [69] model for the ABSA task [70]. Xu et al. trained a BERT model jointly for reading comprehension and ABSA, and showed performance improvement on both tasks [71]. Zeng et al. proposed dynamically weighing or masking the attention weights of the sentence words depending upon its distance from the aspect expression [27]. In this work, we use ABSA models to find the sentiment expressed toward professions in media content.

Standard Occupational Classification taxonomy

The Standard Occupational Classification (SOC) system is a profession taxonomy, maintained by the US Bureau of Labor Statistics [25]. It arranges professions in four tiers: major, minor, broad, and detailed. The detailed tier contains a set of professions, closely related by work. Fig 1 lists all 23 major SOC groups, and shows a portion of the taxonomy’s subtree rooted at Management Occupations major SOC group. As shown in the figure, the profession Governor occurs in the following SOC hierarchy: Management Occupations (major) → Top Executives (minor) → Chief Executives (broad) → Chief Executives (detailed) → Governor (profession). The SOC taxonomy contains 6520 unique professions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Standard Occupational Classification (SOC) system.

The Standard Occupational Classification (SOC) is a 4-tiered profession taxonomy: major, minor, broad, and detailed groups. Detailed groups contain a set of closely-related professions. This figure shows the Management Occupations (major) → Top Executives (minor) → Chief Executives (broad) → Chief Executives (detailed) → Admiral, Chief Financial Officer, …, University President (professions) branch.

https://doi.org/10.1371/journal.pone.0267812.g001

OpenSubtitles

We used the English subset of the OpenSubtitles dataset [72], which contains 135,998 subtitle files corresponding to a variety of media content from the years 1950 to 2017. Each subtitle file is mapped to a unique IMDb title. More than 94% of the IMDb titles are movies or TV show episodes, and the rest are made up of video games, TV shorts, TV mini-series, etc. Subtitle files for IMDb titles released before 1950 are available, but we excluded them because there were very few titles for each year (less than 100) and it would not have been a representative sample for that period’s media content. The subtitle files of our dataset contain around 126 million sentences and 942 million words.

Fig 2 shows some media metadata statistics of our subtitles dataset. The first panel shows the temporal distribution of movie and TV show IMDb titles by each decade. The number of media titles increases with time, with TV episodes surpassing movie releases in the later years. The number of TV episodes is more than three times the number of movies in the most recent decade (2010-2017). The second panel shows the distribution of the top ten genres of the IMDb titles in our dataset. An IMDb title can have multiple genres. Drama and Comedy are the two most common genres, covering more than 80% of the IMDb titles. The third panel shows the distribution of the top ten most common countries where the production company is based. About 68% of the time, the production company was based in the US or the UK.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Descriptive statistics of the OpenSubtitles dataset.

a) Distribution of IMDb title type (movie/TV) by year. b) Distribution of the top ten genres. c) Distribution of the top ten production countries. We use the OpenSubtitles dataset between the years 1950 and 2017.

https://doi.org/10.1371/journal.pone.0267812.g002

Taxonomy creation

We use WordNet synsets to create a searchable profession taxonomy. Fig 4 shows an example of finding new professions from the SOC job titles: Orchestra Conductor and Train Conductor.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Profession taxonomy creation.

We use WordNet synsets to expand the SOC taxonomy and map to SOC major groups. This figure shows how we find new job titles from the SOC job titles, Orchestra Conductor and Train Conductor.

https://doi.org/10.1371/journal.pone.0267812.g004

We use the following method to expand the SOC taxonomy.

Find Substrings: Given a SOC job title, we split it into substrings and join them cumulatively from the end to find candidate job titles. For example, given the job title Chief Executive Officer, we find the new candidate titles—Officer and Executive Officer. We do not split titles that contain conjunctions, prepositions, punctuations, or those that are abbreviations (all letters are in upper case), or if they contain more than five words. In Fig 4, Conductor is a new job title we find from the SOC words, Orchestra Conductor and Train Conductor.

Find WordNet synsets: We find WordNet synsets of the candidate job titles. We retain only those synsets whose semantic class is noun.person or noun.group. We add the hyponym synsets of the retained synsets.

Remove Non-Professional Synsets: We manually check the list of synsets and remove those that do not refer to any profession. As shown in Fig 4, conductor.n.02 denotes some heat or electricity conducting substance, which is not a profession and therefore, it is removed. The final curated list contains 1615 professional synsets.

Synonym Expansion: We collect all synonyms of the professional synsets. These are the new job titles, which we add to the SOC taxonomy. As shown in Fig 4, synsets conductor.n.01 and conductor.n.03 contribute the new job titles: Music Director, Bandleader, Bandmaster and Conductress. The final taxonomy contains 10952 job titles. The number of unigram titles increased from 426 in SOC to 1881 in the expanded taxonomy.

SOC Mapping: We map the job titles in the expanded taxonomy to SOC major groups. This allows us to compare the employment in different SOC professional groups with their frequency in media content. We create the mapping through a semi-automatic process. Given a job title x, we first find the SOC major groups that contain x or has some job title which contains x as a substring. If there exists exactly one such SOC major group, we map x to it. Otherwise, we examine the professional synsets of x and find the mapping manually. Often, different synsets of a job title map to different SOC groups. For example, from Fig 4, we observe that the two synsets of Conductor, conductor.n.01 and conductor.n.03, map to two different SOC major groups, depending upon their respective definitions. We only perform the mapping for the top 500 most occurring job titles in media subtitles. These job titles belong to 562 professional synsets and cover more than 94% of the professional mentions.

Table 1 shows the number of job titles and synsets in the different profession taxonomies. The SOC taxonomy does not contain any synsets and therefore, is not searchable. The expanded taxonomy adds professional synsets and quadruples the number of unigram job titles, making it searchable. The SOC-mapped taxonomy is a subset of the expanded taxonomy which we use to study the relationship of media frequency of professions with their employment trend. Fig 5 shows the structure of the SOC-mapped profession taxonomy. It contains three tiers: SOC major groups, WordNet synsets and job titles. The figure only shows five SOC major groups of the complete taxonomy.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Profession taxonomy sizes.

https://doi.org/10.1371/journal.pone.0267812.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Searchable profession taxonomy.

The SOC-mapped expanded profession taxonomy contains 3 tiers: SOC major groups, WordNet synsets, and job titles. The synsets and unigram job titles make the taxonomy searchable. This figure shows a few nodes of the SOC-mapped taxonomy, which contains 500 job titles and 562 synsets.

https://doi.org/10.1371/journal.pone.0267812.g005

Profession search

We search mentions of the expanded taxonomy’s job titles in the OpenSubtitles corpus. We apply NER and state-of-the-art WSD techniques to prune non-professional mentions. Finally, we train an ABSA model on sentiment-annotated subtitle sentences, and use it to tag professional mentions with their sentiment polarities.

Determining expressed sentiment.

We tag each professional mention with the sentiment (positive, negative, or neutral) expressed towards them in the subtitle sentence (see section Sentiment Analysis). We apply the LCF (Local Context Focus) BERT model to find the targeted sentiment [27]. The LCF model defines a value called semantic relative distance (SRD) for each word in the sentence. SRD is the absolute difference between the word position and the target (professional mention) position in the sentence. The model masks or weighs down the output features of words whose SRD is larger than some threshold, that is, the model reduces the effect of words that are farther away from the profession word in determining the sentiment. The LCF model achieved state-of-the-art accuracy on the Twitter ABSA task [28], recording 75.78 F1 score. It also obtained high scores on the customer reviews dataset [64] of laptops (79.59 F1) and restaurants (81.74 F1). We train the LCF model using sentiment-annotated professional mentions.

We crowdsourced sentiment annotations using Amazon Mechanical Turk. We trained Turkers using expert-labelled examples, and then asked them to annotate the sentiment of five sentences. We selected only those annotators who correctly annotated all five sentences. 52 annotators qualified our test. These annotators then labelled the sentiment of 15,000 professional mentions: two annotations per mention. We retained only those mentions with identical annotations. We were left with 9613 sentiment-annotated professional mentions: 3,316 positive, 1,683 negative, and 4,614 neutral. The dataset contains mentions of 107 professions. To train the LCF model, we divide the dataset into train, validation and test sets. Professions, whose mentions occur in the training set, do not appear in the validation and test set. This prevents the model from overfitting to the target professions of the training set, encouraging it to learn the sentiment from the context and handle mentions of unseen professions. Table 2 shows the distribution of the sentiment classes and the number of professions in each set.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Sentiment-annotated professional mentions dataset.

https://doi.org/10.1371/journal.pone.0267812.t002

We tune the following hyperparameters of the LCF model on the validation set: SRD, architecture type (CDM or CDW), L2-regularization, dropout, embedding dimension, and hidden dimension. The model achieved 87.76% accuracy and 83.22 F1 on the test set. We apply the trained model to find the targeted sentiment of each professional mention of our corpus.

In total, our corpus of professional mentions contains 3,657,827 mentions, covering 4,073 professions. The corpus contains mentions from 133,133 IMDb titles, ranging between the years 1950 to 2017. The top 500 most occurring professions, which have been mapped to SOC major groups (see section Taxonomy Creation), cover more than 94% of the mentions.

Profession frequency

We calculate the media frequency of a profession as the total number of professional mentions (both singular and plural form, for example, advocate and advocates) divided by the total number of n-grams in the subtitles. Here, n equals the number of words in the profession phrase, for example, doctor is a 1-gram, chief executive officer is a 3-gram, etc. We calculate the frequency of SOC major groups by adding the frequencies of professions mapped to it (see section Taxonomy Creation). This frequency measure is motivated by the Google-ngrams study [23]. We calculate the trend of a profession or SOC major group as the Spearman’s rank correlation coefficient [74] of its media frequency against time. A significant positive correlation denotes an increasing trend, and a significant negative correlation implies a decreasing trend over time (α = 0.05).

Fig 6 shows the trend of the 23 major SOC groups, from most positive to most negative. Table 3 lists these groups and their frequency trends. Only 3 SOC groups showed an increasing frequency trend in mentions over time, while 16 SOC groups decreased in frequency over time. The rank correlation was not significant for the remaining 4 SOC groups. These SOC groups contain 500 professions (see section Taxonomy Creation). We analyze the trend of some professions belonging to these SOC groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Spearman’s rank correlation coefficient of the media frequency of SOC groups vs time.

The red-colored bars have positive correlation (increasing trend), and the blue-colored bars have negative correlation (decreasing trend). The grey-colored bars mean that the correlation is not statistically significant (α = 0.05).

https://doi.org/10.1371/journal.pone.0267812.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. SOC major groups with increasing, decreasing or no frequency trend over time.

https://doi.org/10.1371/journal.pone.0267812.t003

Computer and Mathematical Occupations: This SOC group includes mathematicians and computer-related professions. Fig 7a) shows the frequency trend of two of its professions: hacker and programmer. Both occupations showed a positive frequency trend, but mentions of hackers increased more than programmers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Frequency trends of different professions over time in media subtitles.

ADESM = Arts, Design, Entertainment, Sports, and Media. FPSR = Food Preparation and Serving Related.

https://doi.org/10.1371/journal.pone.0267812.g007

Life, Physical, and Social Science Occupations: This SOC group includes archaeologists, astronauts, biologists, chemists, geologists, etc. Almost all its professions showed an increasing frequency trend over time. Fig 7b) shows the trend of four occupations: geologist, biologist, anthropologist, and economist. Mentions of geologists and biologists are consistently more frequent than economists and anthropologists.

Arts, Design, Entertainment, Sports, and Media Occupations: This SOC group includes entertainment, sports, music, and media-related professions. Fig 7c), 7d), 7e) and 7f) show the frequency trends of some of its professions. Mentions of actor dominate actress mentions in media content. The word actor can be used as a gender-neutral term, explaining part of this trend. The frequency of sports-related professions dipped in the 1960s but has increased since then. Pianist mentions decreased, whereas mentions of bass players, guitarists, and drummers increased over time. Mentions of journalists and reporters are more frequent than correspondents and columnists.

Legal Occupations: Legal occupations include lawyers, judges, attorneys, prosecutors, etc. Fig 7g) shows the frequency trend of defense attorneys and prosecutors. Mentions of prosecutors (a lawyer who conducts a case against a defendant) are more frequent than defense attorneys (a lawyer who defends the client against criminal charges).

Food Preparation and Serving Related Occupations: This SOC group includes professions related to food serving and food preparation. Fig 7h), 7i) and 7j) show the frequency trends of some of its occupations. Mentions of waiters are more frequent than waitresses overall. Similar to actors, waiters can refer to either gender, so this trend is not surprising. However, mentions of waiters have decreased over time, whereas mentions of waitresses have increased. Gendered professional mentions like barman and barmaid are less frequent than the gender-neutral term: bartender. Mentions of cooks were more common than chefs, but the latter became more frequent in media content in the 2010s.

Healthcare Support Occupations: This SOC group includes healthcare assistants, nursing aides, massage therapists, etc. Fig 7k) shows the frequency trend of two gendered professions: masseur (male) and masseuse (female). The frequency of masseurs has significantly decreased over time after it peaked around 1970. Mentions of masseuses have become more common than masseurs. The frequency of the gender-neutral term, massage therapist, has increased over time.

Business and Financial Operations Occupations: This SOC group includes accountants, contractors, auditors, etc. Fig 7l) shows the frequency trend of accountants and auditors. Their frequencies have mostly remained steady over time. Accountants are more frequent than auditors.

Personal Care and Service Occupations: This SOC group includes barbers, valets, nannies, ushers, etc. Fig 8a) shows the frequency trend of three child-care-related professions: nanny, babysitter, and governess. The mention frequency of nannies and babysitters has increased over time. The word governess is an archaic term, and its usage has declined.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Frequency trends of different professions over time in media subtitles.

CSS = Community and Social Service.

https://doi.org/10.1371/journal.pone.0267812.g008

Installation, Maintenance, and Repair Occupations: This SOC group includes mechanics, electricians, locksmiths, etc. Fig 8b) shows the frequency trend of electricians and mechanics. The mentions of both these professions have decreased in media subtitles.

Farming, Fishing, and Forestry Occupations: This SOC group includes farmers, shepherds, herders, fishermen, etc. Fig 8c) shows the frequency trend of farmers, fishermen, and hunters. Mentions of farmers and fishermen have decreased, whereas mentions of hunters have increased over time.

Architecture and Engineering Occupations: This SOC group includes architects, designers, engineers, surveyors, etc. Fig 8d) shows the frequency trend of engineers and architects. The frequency of architect mentions has remained steady over time, whereas mentions of engineers have diminished.

Sales and Related Occupations: This SOC group includes sales and real-estate-related professions. Fig 8e) and 8f) show the frequency trends of some of its occupations. Mentions of retailers, distributors, vendors, and brokers have increased over time. The increase in frequency is even more prevalent in real-estate jobs: estate agent, realtor, and real estate agent.

Management Occupations: This SOC group includes administrative professions and occupations related to the management of educational institutions. Fig 9g) and 9h) show the frequency trend of some management occupations. Mentions of both congressman and congresswoman have increased over time. Although congressmen are mentioned more than congresswomen, the frequency of congresswomen has increased at a higher rate (not discernible from the graph). Mentions of senators, governors, and mayors have decreased over time. The frequency of headmistress mentions has increased, whereas headmaster mentions have decreased in media content. The frequency of the gender-neutral term, principal, has increased.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Frequency trends of different professions over time in media subtitles.

HPT = Healthcare Practitioners and Technical. TMM = Transportation and Material Moving. BGCM = Building and Grounds Cleaning and Maintenance.

https://doi.org/10.1371/journal.pone.0267812.g009

Educational Instruction and Library Occupations: This SOC group includes teachers, professors, librarians, etc. Fig 8i) shows their frequency trends. Professor mentions have decreased greatly over time, whereas the frequency of teacher and librarian mentions have increased slightly.

Construction and Extraction Occupations: This SOC group includes builders, carpenters, masons, miners, etc. Fig 8j) shows their frequency trends. The frequency of masons has remained steady, but miner and carpenter mentions have reduced over time.

Community and Social Service Occupations: This SOC group includes religious and social workers. Fig 8k) and 8l) shows the frequency trend of some religious occupations. The mention frequency of bishops and reverends has decreased over time, whereas the frequency of pastors and deacons has increased. Mentions of imams, lamas, and rabbis have also increased over time, but the frequency of priests has decreased.

Healthcare Practitioners and Technical Occupations: This SOC group includes healthcare-related professions. Fig 9a) and 9b) show the frequency trend of some of its professions. Mentions of healthcare specialists like cardiologists, chiropractors, gynecologists, and neurologists have increased over time, but the frequency of the generic term, doctor, has decreased.

Protective Service Occupations: This SOC group includes law enforcement and protective service workers. Fig 9c) and 9d) show the frequency trend of some of its professions. Mentions of policemen have decreased over time but are still much higher than mentions of policewomen, which have remained steady. Guard occupations like bodyguards, prison guards, and security guards have become more frequent in media content.

Production Occupations: Production occupations include artisans, cobblers, jewelers, millers, etc. Fig 9e) shows the frequency trend of some of its professions: jeweler, shoemaker, and tailor. The mention frequency of all three professions has decreased over time.

Transportation and Material Moving Occupations: This SOC group includes professions related to the transportation of goods and people. Fig 9f), 9g) and 9h) show frequency trends of different transportation-related professions. Mentions of bus drivers increased, cab drivers remained steady, and truck drivers decreased over time. Mentions of sailors decreased more rapidly than pilots. The frequency of gendered professional terms like steward and stewardess decreased over time. The frequency of the gender-neutral term, flight attendant, increased.

Office and Administrative Support Occupations: This SOC group includes clerks, receptionists, tellers, notaries, etc. Fig 9i) shows the frequency trend of some of its professions: clerk, secretary, and bookkeeper. The frequency of all three professions decreased over time.

Building, Grounds Cleaning and Maintenance Occupations: This SOC group includes construction workers, maids, chamberlains, janitors, etc. Fig 9j) shows the frequency trend of some of its professions: gardener, sweeper, and caretaker. Mentions of all three professions decreased over time in media content.

Military Specific occupations: Military occupations include army, naval and air-force-related professions. Fig 9k) shows the frequency trend of some of its professions. Mentions of field marshal (army) and admiral (navy) decreased over time. The frequency of air marshal mentions increased (not discernible from the graph).

The frequency trend of the SOC group does not reflect the frequency trend of all its professions. For example, Fig 8e) and 8f) show increasing trends for many sales-related professions, but the overall frequency of the Sales Occupations SOC group decreased. A large proportion of Sales Occupations mentions are comprised of bankers and cashiers, whose frequencies have decreased.

Profession sentiment

We find the sentiment expressed toward each professional mention in our dataset using the LCF model (see section Profession Search). The computed sentiment can take the following values: positive, negative, or neutral. We call all mentions tagged with non-neutral sentiment as opinionated mentions. We represent the sentiment expressed towards a profession or a major SOC group as the number of positive sentiment mentions divided by the total number of opinionated mentions. We find the sentiment trend of professions by calculating the Spearman’s rank correlation [74] between the proportion of positive sentiment mentions and time.

Fig 10 shows the proportion of positive sentiment mentions of the 23 major SOC groups. From the figure, we observe that the proportion is always greater than 0.5. Therefore, the number of positive sentiment mentions is greater than the number of negative sentiment mentions for all the SOC groups. This might be because of an inherent positive bias in the sentiment of media narratives. Architects and engineers are talked about most positively, and sales-related occupations are talked about most negatively. STEM professions are generally expressed in positive sentiment. Four out of the top ten SOC groups ordered by the proportion of positive sentiment contain STEM professions: Architecture and Engineering; Computer and Mathematical; Life, Physical, and Social Science; and Healthcare Practitioners and Technical Occupations. Professions involving manual labor are generally expressed with negative sentiment. Four out of the bottom ten SOC groups contain blue-collar jobs: Building, Grounds Cleaning and Maintenance; Production; Farming, Fishing, and Forestry; and Construction and Extraction Occupations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Proportion of positive sentiment mentions in opinionated mentions of the SOC groups.

https://doi.org/10.1371/journal.pone.0267812.g010

We analyze the proportion of positive sentiment mentions for individual professions. Fig 11 shows the top ten professions with the highest proportion of positive sentiment mentions and the bottom seven professions with the lowest proportion of positive sentiment mentions from the list of most occurring top 100 professions in our subtitle corpus. More than half of the opinionated mentions of police and cops are negative. Religious workers like monks and nuns are talked about more negatively than positively. More than four-fifths of the mentions of singers, musicians, and dancers are positive. STEM professions like astronauts and engineers are also mentioned positively in media content.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 11. Top 10 positive-sentiment and top 7 negative-sentiment professions.

The blue-colored professions have the top 10 highest proportion of positive sentiment mentions. The red-colored professions have the top 10 highest proportion of negative sentiment mentions.

https://doi.org/10.1371/journal.pone.0267812.g011

We also study the sentiment trend of professions in the analyzed media content. Fig 12 shows the Spearman’s rank correlation coefficient of the proportion of positive-sentiment mentions over time for some of the professions. The figure lists the top six professions with the highest correlation and the bottom ten professions with the lowest correlation from the list of most occurring top 100 professions in our subtitle corpus. The sentiment expressed towards therapists, spies, rangers, and detectives in media content trends toward becoming more positive over time, whereas the sentiment expressed toward doctors, lawyers, professors, and scientists trends more negative. Lawyer mentions have the second-lowest sentiment correlation over time, behind doctors. Astronauts not only have the highest proportion of positive sentiment mentions overall (see Fig 11), but also one of the most positive sentiment trends.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 12. Professions with the top increasing and top decreasing sentiment trends.

The blue-colored professions have the top 6 most positive Spearman’s rank correlation of proportion of positive mentions vs time. Therefore, the proportion of positive sentiment mentions shows increasing trend over time for these professions. The red-colored professions have the top 10 most negative correlation and their proportion of negative sentiment mentions is increasing over time.

https://doi.org/10.1371/journal.pone.0267812.g012

Media attributes

We have observed that both the frequency of professional mentions and the sentiment expressed towards them change over time. In this section, we study the relationship of the following media attributes: year, genre, title type, and country of production (see section OpenSubtitles) with the observed media frequency and sentiment trends of professions and SOC major groups.

We perform a regression analysis to analyze the effect of media attributes. The response is the frequency or sentiment of the profession or the SOC major group. The predictors are year, genre, title type, and country of production. All predictors are categorical variables except for year, which is numeric. Genre and country are multi-valued. An IMDb title can belong to multiple genres, and its production can take place in multiple countries. Therefore, we create a categorical variable for each possible genre and country, taking values 0 or 1. We set it to 1 if the IMDb title belongs to the corresponding genre or its production takes place in the corresponding country. We ignore media attribute configurations for which the number of IMDb titles is less than 30. We use logistic regression because both frequency and sentiment are proportions, bounded between 0 and 1 (defined in sections Profession Frequency and Profession Sentiment). We use a generalized linear model with the binomial family and logit link. We provide the total number of ngrams and opinionated mentions (defined in section Profession Sentiment) as prior weights to the model to specify the number of trials when the response is frequency or sentiment, respectively.

Figs 13 and 14 show the significant predictors (α = 0.05) when the response is the frequency of professions and SOC groups, respectively, in the form of a heatmap. The color of the cells indicates the sign of the coefficient (blue = +, red = −) and the intensity of the color denotes its magnitude. The white cells mean that the predictor is not significant. We observe some interesting relationships between the frequency of professional mentions and media attributes. The frequency of actors increases, but the frequency of actresses decreases when the genre is adventure, documentary, or thriller. The reverse is true when the genre is romance. Mentions of lawyers increase in crime, drama, and mystery genre media content. The frequency of lawyers and attorneys increase, and the frequency of prosecutors decreases when the country of production is the United States. United States-produced movies and TV shows mention cops and sheriffs more than inspectors and police. The opposite is true for United Kingdom-produced titles. The frequency of detectives and spies increases in mystery genre titles. Comedy, reality-TV, and music genres increase the frequency of dancers, singers, and artists. The frequency of doctors, nurses, and surgeons in movies is higher than in TV shows. In science-fiction and family media titles, the frequency of doctors increases, and the frequency of nurses and surgeons decreases. Documentary titles increase the frequency of reporters and journalists. Mentions of engineers and scientists increase when the genre is science-fiction or documentary and decreases in comedy and fantasy genres. The frequency of teachers and professors decreases in action and adventure genres. Movies mention teachers more than TV shows, but the opposite holds for professors. Mentions of senators, mayors, and presidents increases in news and thriller genres. The frequency of lieutenants and soldiers increases in action and war media titles and decreases when the genre is fantasy and romance.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 13. Heatmap of coefficient values of media attributes in predicting profession frequency.

The color denotes the sign of the coefficient (blue = +, red = −) and the intensity of the color is proportional to the magnitude of the coefficient. The blank cells indicate that the media attribute is not a significant predictor for the frequency of the corresponding profession.

https://doi.org/10.1371/journal.pone.0267812.g013

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 14. Heatmap of coefficient values of media attributes in predicting SOC group frequency.

The color denotes the sign of the coefficient (blue = +, red = −) and the intensity of the color is proportional to the magnitude of the coefficient. The blank cells indicate that the media attribute is not a significant predictor for the frequency of the corresponding SOC group.

https://doi.org/10.1371/journal.pone.0267812.g014

Fig 14 shows the coefficient heatmap of media attributes when the response is the frequency of SOC groups. We highlight some relationships between SOC frequency and media attributes. The frequency of Management, Business, and Financial Operations occupations increases in biography and news genre media titles. Documentaries and science-fiction movies and TV shows frequently mention STEM professions like Computer, Mathematical, Architecture, Engineering, Life, Physical, and Social Science occupations. Mentions of Community and Social Service occupations decrease in reality TV shows, sports, and family movies. The frequency of Legal occupations increases in news genre titles. The frequency of Arts, Design, Entertainment, Sports and Media occupations increases in music, sports, game show, and biography genre titles. Mentions of Healthcare Practitioners increase in news and drama genres and decrease in musicals. Food Preparation and Serving related professions occur highly in reality TV shows and decrease in music and adventure movies. The frequency of manual labor jobs like Construction, Extraction, Production, Building, Grounds Cleaning, and Maintenance occupations decreases when the country of production is the United States. Mentions of Military occupations increase in war and action movies and decrease in comedy and family shows.

We study the effect of media attributes on the proportion of positive sentiment mentions of professions and SOC Groups. Unlike frequency, the number of significant predictors are fewer. Therefore, instead of a heatmap, we tabulate the significant media attributes with the most positive and negative coefficients (α = 0.05) in Table 4. The proportion of positive sentiments expressed towards marshalls, mayors, professors, and prosecutors increases when the genre is mystery. The proportion of negative sentiment increases for congressmen, priests, and prosecutors in crime genre media titles. The average sentiment of SOC groups containing STEM occupations like Computer, Mathematical, Life, Physical, and Social Science occupations becomes more positive in documentaries. Movies express more negative sentiment than TV shows towards Legal occupations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Effect of media attributes on the sentiment of professions and SOC groups.

https://doi.org/10.1371/journal.pone.0267812.t004

Employment

We analyzed the frequency and sentiment trends of different professions and SOC groups in media content. We also studied the effect of media attributes on these representations. However, our analysis has been limited largely to the entertainment media domain, by using NLP on the subtitles of media content. In this section, we study the correlation between media frequency and real-world events, namely employment figures. We obtained the employment data of SOC major groups from the Occupational Employment Statistics survey (OES). The survey does not provide employment numbers for individual professions. Therefore, we only conduct our analysis on SOC groups. We calculate the Spearman’s rank correlation [74] between the media frequency of the SOC group and the proportion of the working population employed in any of the professions of the SOC group. We compute the correlation for the period 1999-2017. The employment data for the earlier years is not available.

Fig 15 shows the correlation between media frequency and employment for the SOC groups, from most positive to most negative. The correlation is positive for 14 out of the 22 SOC groups (64%) and negative for the rest. Therefore, the trend of media frequency correlates with the employment trend for most SOC groups. The number of media mentions of the SOC group increases as more people are employed in its professions, and decreases as people move away to other occupations. The correlation is significantly negative for only two SOC groups: Healthcare Practitioners and Social Service occupations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 15. Spearman’s rank correlation coefficient of the media frequency of SOC groups with employment.

The red-colored bars have positive correlation: media frequency and employment have the same trend. The blue-colored bars have negative correlation: media frequency and employment have opposite trends. The grey-colored bars mean that the correlation is not statistically significant (α = 0.05).

https://doi.org/10.1371/journal.pone.0267812.g015

Findings from frequency analysis

We observed that gender-neutral terms like massage therapists and flight-attendants are becoming more frequent than their gendered counterparts. We suggest that this trend is due to the increased awareness stemming from conversations about gender-neutral terms among the youth, parents, and the LGBTQ+ community [76, 77]. Analyzing explicit gendered profession terms, we observed that the frequency of some female job titles such as waitresses, congresswomen, and policewomen has either increased or remained steady relative to the corresponding male job titles (i.e., waiters, congressmen, and policemen). Equal opportunity legislation, women’s rights movements [78], and constitutional amendments for women’s suffrage [79] have improved the representation of women in police and government. This may have driven a similar trend of increasing female representation in media content. While there is a change in trend between the explicitly gendered job titles over the years, the overall frequency of most male job titles exceeds their female counterparts.

The frequency of STEM, sports, arts, and design occupations has increased, whereas the frequency of construction, farming, and manual labor jobs has decreased. Improvements in cinematography and visual effects technology, space travel, the emergence of the world wide web, and sports documentaries have contributed to the increased fascination of science and sports in media [80, 81]. These advances and interests potentially explain the increase in the frequency of STEM-related and sports portrayals on the big screen. The frequency of specialized professions like cardiologists, gynecologists, and neurologists has increased, but the frequency of generic terms like doctors and nurses has decreased. The evolution of medicine and research, students’ preference for higher-paying specialties [82], and the emergence of novel diseases have increased the demand for specialized professions over general ones in the medical field [83]. We posit that as new and specialized professions become popular career choices and essential in catering to the needs of the continuously evolving society, they also attract the interest of media creators.

Thus far, we have discussed the frequency trends of individual professions. It is important to note that this does not always reflect the overall trend of the subsuming SOC group. Sales-related occupations like retailers, vendors, brokers, and realtors show an increase in frequency, but the aggregate frequency of the associated SOC group shows a decreasing trend. Technology improvements in computerized financial transactions (barcode scanning, automatic teller machines, and mobile banking) and customers’ preference for self-service checkouts probably contributed to the decrease in employment of bankers and cashiers [84, 85], or at least their explicit mentions in media narratives.

Findings from sentiment analysis

STEM occupations are favorably (i.e., positive sentiment) mentioned in the subtitles corpus we analyzed. However, the sentiment expressed toward professions that predominantly involve manual labor—the so-called “blue-collar” jobs—is largely negative. Overall, police, monks, and nuns are mentioned negatively, whereas musicians and engineers are mentioned positively.

The sentiment expressed toward therapists, astronauts, and detectives was observed to trend more positive over time, while the sentiment toward doctors, lawyers, professors, and presidents, was more negative. The decreasing trend of positive sentiment toward lawyers is consistent with Asimow’s work [16] that showed the decline in public opinion of lawyers. This may be one of the explanatory factors behind the increasingly negative portrayal of lawyers in movies. In contrast, the negative sentiment expressed towards doctors, scientists, and police that we found in our media corpus does not appear to agree with the findings of public polls. Several studies over the years [86–88] show that people usually trust these professions. We note that public opinion also depends on several social and demographic factors such as political leanings and the level of educational attainment [87, 88]. Factors like the popular stereotype of the “mad scientist” (someone who is more concerned about their scientific findings than human welfare) [18] in movies, grievances of the health care system [89], and general distrust towards people and institutions in positions of authority, might have contributed towards these negative portrayals in media.

We have made several hypotheses to explain the observed frequency and sentiment trends and require controlled experiments to verify these claims. It is challenging to design such studies because we are trying to relate media narratives to real-world events—two very different domains. As of now, we are unsure regarding what types of experiments are required, but we suggest some ideas. We can survey scriptwriters and movie directors to find what societal events inspired their stories and creative decisions. We can also try to map change points in media trends to real-world events to explain their possible cascading effects on media narratives and genres. We leave this task for future research.

Findings from media attribute analysis

Media attributes like genre, country of production, and content type affect the frequency and sentiment trends of professions in our subtitle corpus. Genre, in particular, seems to be a good predictor of the type of profession mentioned in the subtitles. For example, science-fiction movies often mention engineers and scientists, action and war genres mention lieutenants and soldiers, and mystery titles contain detectives and spies. Adventure/thrillers contain more references to male actors than female actors. However, the opposite is true for romantic movies. Consistent with findings in previous work [90], this suggests that gender bias is prevalent in these media genres.

In examining the country of production, we found that sheriffs are mentioned more than inspectors in titles produced in the US, and the opposite holds for movies produced in the UK, reflecting the differences in English usage, and law enforcement structures in these countries. Regarding the type of content, we did not find significant differences in the mentions of job titles between movies and TV shows in our subtitle corpus.

Findings from employment analysis

Interestingly frequency of job titles in media correlates with real-world employment statistics of corresponding professions. We observed a significant positive correlation between media frequency and employment trends of most SOC groups. Professions that typically employ more people were also more frequently mentioned in media content. We note that these correlation findings do not imply causality; a question that requires further systematic study.

Limitations

There are several limitations of the present work. First, subtitles provide only a limited view of the complete movie or TV show. Features, such as the character’s behavior on screen, their interactions with other people, and their character arc can only be studied by carefully viewing the entire movie or TV episode. Therefore, our representation measures of frequency and sentiment of job titles in subtitles do not capture all aspects of their media portrayal. Second, our frequency and sentiment analyses do not control for the frequency or sentiment expressed toward these professions in everyday language. For example, the frequency of policewomen might have been less than the frequency of policemen in media subtitles because that is how their usage evolved in literature. A possible solution is to use the corresponding Google ngram frequencies [23] of the professional words as covariates in our analysis. Third, our study has primarily focused on the representation of professions in the US and the UK (see the distribution of production country in section OpenSubtitles). One needs to adapt our proposed taxonomy and models for different cultures and languages. Our profession taxonomy does not contain all job titles in other countries. Depending on the application of this taxonomy, one may have to add professions specific to a local region for a comprehensive study.